WO2020154962A1 - 一种目标可信度确定方法、一种目标识别方法、系统、车辆及存储介质 - Google Patents

一种目标可信度确定方法、一种目标识别方法、系统、车辆及存储介质 Download PDF

Info

Publication number
WO2020154962A1
WO2020154962A1 PCT/CN2019/073964 CN2019073964W WO2020154962A1 WO 2020154962 A1 WO2020154962 A1 WO 2020154962A1 CN 2019073964 W CN2019073964 W CN 2019073964W WO 2020154962 A1 WO2020154962 A1 WO 2020154962A1
Authority
WO
WIPO (PCT)
Prior art keywords
target
detection target
detection
preset
information
Prior art date
Application number
PCT/CN2019/073964
Other languages
English (en)
French (fr)
Inventor
陆新飞
李怡强
陈雷
Original Assignee
深圳市大疆创新科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市大疆创新科技有限公司 filed Critical 深圳市大疆创新科技有限公司
Priority to PCT/CN2019/073964 priority Critical patent/WO2020154962A1/zh
Priority to CN201980005757.0A priority patent/CN111406224A/zh
Publication of WO2020154962A1 publication Critical patent/WO2020154962A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/02—Systems using reflection of radio waves, e.g. primary radar systems; Analogous systems
    • G01S13/50—Systems of measurement based on relative movement of target
    • G01S13/52—Discriminating between fixed and moving objects or between objects moving at different speeds
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/91—Radar or analogous systems specially adapted for specific applications for traffic control
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/93—Radar or analogous systems specially adapted for specific applications for anti-collision purposes
    • G01S13/931—Radar or analogous systems specially adapted for specific applications for anti-collision purposes of land vehicles

Definitions

  • the embodiments of the present application relate to the field of unmanned driving technology, and in particular, to a method for determining the credibility of a target, a method for identifying a target, a system, a radar, a movable platform, and a storage medium.
  • assisted driving and autonomous driving have become current research hotspots.
  • the recognition of targets is essential to realize unmanned driving.
  • functions such as deceleration and avoidance, emergency stop, obstacle bypass, lane change, and automatic stop by station can be implemented according to the specific situation.
  • the image of the target is mainly collected by the vision sensor to identify the target.
  • the vision sensor does not have all-weather and all-weather characteristics when recognizing targets. For example, in weak light, rain, snow, and foggy weather, the image collection effect of the visual sensor is not good, which will significantly reduce the effect of the visual sensor on target recognition. At this time, if you rely solely on visual sensors, a large number of scenarios where automatic driving fails may occur.
  • millimeter-wave radar has the advantages of all-weather and all-weather when working, but related technologies have not yet achieved effective target recognition through millimeter-wave radar.
  • the embodiments of the present application provide a method for determining the credibility of a target, a target recognition method, a system, a radar, a movable platform, and a storage medium.
  • an embodiment of the present application provides a method for determining target credibility, and the method includes:
  • the reliability of the type of the detection target is determined.
  • an embodiment of the present application provides a target recognition method, which includes:
  • the detection target is a specific target type.
  • an embodiment of the present application provides a target credibility determination system.
  • the system includes one or more processors that work together or individually, and the processors are configured to perform the following operations:
  • the reliability of the type of the detection target is determined.
  • an embodiment of the present application provides a target recognition system.
  • the system includes one or more processors that work together or individually, and the processors are configured to perform the following operations:
  • the detection target is a specific target type.
  • an embodiment of the present application provides a radar, which includes:
  • An antenna which is used to obtain echo signals
  • the processor is in communication connection with the antenna, and is configured to execute the method described in any one of the technical solutions of the first aspect and/or the second aspect of the present application.
  • an embodiment of the present application provides a movable platform, which includes:
  • the power system is installed on the body to provide power
  • an embodiment of the present application provides a computer-readable storage medium, characterized in that a computer program is stored thereon, and the computer program is executed by a processor to achieve the objective provided by the technical solution of the first aspect of the present application The steps of the method for determining the credibility and/or the target recognition method provided by the technical solution of the second aspect of the present application.
  • the detection information of the target and the information of the Doppler unit around the target are acquired, and the target is identified according to the acquired information. It can effectively identify targets under all-time and all-weather conditions, and improve the stability and safety of automatic driving/assisted driving.
  • FIG. 1 shows a schematic diagram of an application scenario provided by an embodiment of the present application
  • FIG. 2 shows a schematic flowchart of a method for determining target credibility according to an embodiment of the present application
  • FIG. 3 shows a schematic flowchart of a method for determining target credibility according to an embodiment of the present application
  • FIG. 4 shows a schematic flowchart of a method for determining target credibility according to an embodiment of the present application
  • FIG. 5 shows a schematic flowchart of a target recognition method provided by an embodiment of the present application
  • FIG. 6 shows a schematic flowchart of a target recognition method provided by another embodiment of the present application.
  • FIG. 7 shows a schematic flowchart of a target recognition method provided by another embodiment of the present application.
  • FIG. 8A shows a Doppler-range plane image provided by a specific embodiment of the present application
  • FIG. 8B shows an algorithm implementation scheme of a specific embodiment of the present application
  • FIG. 9 shows a schematic structural diagram of a target credibility determination system according to an embodiment of the present application.
  • FIG. 10 shows a schematic structural diagram of a target recognition system provided by an embodiment of the present application.
  • FIG. 11 shows a schematic structural diagram of a radar provided by an embodiment of the present application.
  • FIG. 12 shows a schematic structural diagram of a movable platform provided by an embodiment of the present application.
  • FIG. 13 shows a schematic structural diagram of a movable platform provided by an embodiment of the present application.
  • a target credibility determination method In order to improve the safety of the automatic driving/assisted driving system, embodiments of the present application propose a target credibility determination method, target recognition method, system, mobile platform, and storage medium.
  • the target is identified by at least one of the above-mentioned sensors mounted on the movable platform.
  • the target can be people, animals, trees, vehicles, street signs, fences, drones, etc.
  • the target recognition method can be used to determine the type of target, determine the target's trajectory, plan the trajectory of the movable platform, and the operation of the movable platform.
  • the credibility is an evaluation index that measures the credibility of the type of detection target.
  • the information of the detection target returned in the echo signal is acquired, and the reliability of the type of the detection target is generated according to the acquired information of the detection target.
  • the higher the credibility the higher the credibility to prove the type of detection target; the lower the credibility, the lower the credibility to prove the type of detection target.
  • the type of detection target is finally determined.
  • the detection target by detecting the same detection target multiple times, and according to the current detection result of the detection target and the last detection result of the detection target, it is determined that the detection target is a specific target type, such as a pedestrian target, no Human and machine goals, etc.
  • the track of the target can also be recorded and predicted, so as to track the target.
  • the movement trajectory of the movable platform can be planned according to the current trajectory of the target. For example, deceleration and avoidance, emergency stop, obstacle bypass, lane change, automatic stop by station, etc.
  • Figure 1 provides an example of an application scenario provided by an embodiment of the present application.
  • the movable platform is an unmanned vehicle 10
  • the movable platform includes a vehicle body 101 and a radar 102.
  • the radar 102 is installed on the vehicle body 101.
  • the radar can be installed on a movable platform, such as robots, unmanned aerial vehicles, unmanned vehicles, ordinary vehicles, VR glasses, AR glasses, etc. Taking the installation of a radar in an unmanned vehicle as an example, the radar can be integrated in one or more locations of the vehicle, or a device installed on the vehicle, such as on-board equipment, etc. There is no restriction on this. Among them, the radar may be a millimeter wave radar or other types of radar sensors, and there is no restriction on this.
  • the radar includes at least an antenna, and the antenna is used to receive echo signals.
  • the radar 102 moves with the movement of the unmanned vehicle 10 to detect the target to be detected, so as to obtain data for target credibility determination and target recognition-to obtain the detection information of the detected target and the Detect the information of the Doppler unit around the target, and determine the reliability of the type of the detection target according to the detection information of the detection target and the information of the Doppler unit around the detection target;
  • the current detection result of the detection target and the last credibility of the detection target determine the current credibility of the detection target; according to the current credibility, it is determined that the detection target is a specific target type.
  • the radar 102 used for target credibility determination and target recognition in this embodiment may be a millimeter microwave radar.
  • FIG. 2 is a schematic flowchart of a method for determining target credibility according to an embodiment of the application. As shown in Figure 2, the method may include:
  • the execution body can be a radar or a processor installed on a movable platform.
  • the following takes the execution subject as the radar as an example to introduce the target credibility determination method in detail.
  • the detection information of the detection target includes at least one of the following: distance, scattering intensity, speed information, angle information, and observation energy.
  • the distance information includes the radial distance of the detection target relative to the radar;
  • the scattering intensity information includes the echo intensity of the scattered echo generated by the detection target under the radar wave;
  • the velocity information includes the detection target determined by the Doppler frequency shift Relative to the radial velocity data of the radar;
  • the angle information is the angle of the target object compared to the radar;
  • the observation energy is the energy of the echo signal.
  • the radar obtains the detection information of the detection target by emitting electromagnetic waves and receiving feedback echo signals.
  • the echo signal is a frequency modulated continuous wave (FMCW), for example, a fast scan waveform, a triangle wave, or a sawtooth wave.
  • FMCW frequency modulated continuous wave
  • the information of the Doppler unit around the detection target includes at least one of the following: Doppler-range Doppler units around the detection target on the detection target obtained when the radar detects the target The Doppler energy of each Doppler unit around the target and the number of Doppler units whose Doppler energy is greater than the preset Doppler energy threshold. In this way, the number of Doppler cells whose Doppler energy is greater than the preset Doppler energy threshold is determined.
  • acquiring the information of the Doppler unit around the detection target may include, but is not limited to: acquiring a Doppler-range plane image by radar, the Doppler-range plane image including multiple Doppler units around the target ; Compare the energy value of the Doppler unit with the preset Doppler energy threshold, and record the number of Doppler units whose Doppler energy is greater than the preset Doppler energy threshold.
  • the Doppler unit whose Doppler energy is greater than the preset Doppler energy threshold is added to the Doppler unit set; otherwise, the Doppler unit is discarded.
  • the preset Doppler energy threshold may be a preset fixed value, or it may be an energy value of a reference Doppler unit.
  • the Doppler energy threshold can be determined by the intensity of the noise signal of the radar and the Doppler energy corresponding to different types of targets. By counting the average energy of the noise, the average energy of the noise and the constant Pn are selected as the Doppler energy threshold.
  • the constant Pn is the energy information of the target collected by the radar. Among them, Pn is obtained through multiple collection and training of different types of targets.
  • the detection information of the detection target includes at least one of the following: distance, scattering intensity, and velocity information; and the information of the Doppler unit around the detection target includes at least one of the following: The Doppler energy of each Doppler unit and the number of Doppler units whose Doppler energy is greater than the preset Doppler energy threshold are taken as an example for description.
  • acquiring the detection information of the detection target and the information of the Doppler unit around the detection target specifically includes:
  • the information of the Doppler unit around the detection target is acquired. That is, at the same time or after the detection target is detected, the detection information of the detection target and the information of the Doppler unit around the detection target are acquired.
  • first obtain the detection information of the detection target and if the detection information of the detection target meets the first preset condition, obtain the information of the Doppler unit around the detection target; and/or, if the detection information
  • the detection information of the target does not satisfy the first preset condition, and then the information of Doppler units around the detection target is no longer acquired.
  • a judgment is made first, and only when the first preset condition is met, the information of the Doppler unit around the detection target is acquired. Thereby, the number of times of acquiring the information of the Doppler unit around the detection target is reduced, the amount of calculation is reduced, and the calculation speed of the system is improved.
  • judging whether the detection information meets the first preset condition specifically includes but is not limited to the following method: judging whether the specific parameters included in the detection information are within the corresponding threshold range. For example, it is determined whether the scattering intensity is within a preset scattering intensity threshold range, whether the speed information is within a preset speed threshold range, and so on. The following are examples of the following three scenarios:
  • the detection information includes scattering intensity information, and if the scattering intensity is within a predetermined scattering intensity threshold range, the detection information of the detection target meets the first preset condition, and the Doppler unit around the detection target is acquired Information. In some embodiments, if the scattering intensity is not within the preset scattering intensity threshold range, the detection information of the detection target does not meet the first preset condition, and no acquisition of Doppler units around the detection target is performed. information.
  • the detection information includes speed information, and if the speed information is within a preset speed threshold range, the detection information of the detection target meets the first preset condition, and the acquisition of Doppler units around the detection target is performed. information. In some embodiments, if the speed information is not within the preset speed threshold range, the detection information of the detection target does not meet the first preset condition, and no information about Doppler units around the detection target is acquired. .
  • the detection information includes scattering intensity information and speed information. If the scattering intensity is within a preset scattering intensity threshold range and the speed information is within a preset speed threshold range, the detection information of the detection target satisfies The first preset condition is to acquire information of Doppler units around the detection target. That is, only when both are met, the detection information of the detection target meets the first preset condition, and then the information of the Doppler unit around the detection target is acquired. In some embodiments, if the scattering intensity is not within the preset scattering intensity threshold range, the velocity information is not within the preset velocity threshold range, and the detection information of the detection target does not meet the first preset condition, and To obtain the Doppler unit information around the detection target.
  • determining the scattering intensity threshold range specifically includes but is not limited to the following methods:
  • the distance between the radar and the radar is different, and the detection information contained in the echo signal obtained by the radar is also different.
  • the radar performs multiple detections on the same target to obtain the scattering intensity of the targets located at different distances.
  • the obtained scattering intensities of targets located at different distances are averaged in advance to improve the accuracy of the process of determining the credibility of the target.
  • the radar detects the targets multiple times to obtain the scattering intensity of the targets located at different distances.
  • the scattering intensity of pedestrian targets at different distances is taken as an example.
  • the radar obtains the scattering intensity of pedestrian targets at different distances by detecting pedestrian targets. After that, note down the distance information and the scattering intensity information corresponding to the distance information at the same time.
  • the obtained scattering intensity of pedestrian targets located at different distances is averaged to form the average scattering intensity and distance information change curve, or the average scattering intensity of the pedestrian target and the distance information look-up table.
  • the scattering intensity threshold is determined according to the average value of the scattering intensity and the distance information change curve.
  • the respective scattering intensity threshold ranges are determined for different types of targets.
  • the scattering intensity threshold is also determined by the noise signal intensity of the radar. By counting the average intensity of the noise, the average intensity of the noise and the obtained scattering intensity of the target at different distances are added as the threshold of the scattering intensity.
  • the radar installed on the mobile platform detects an external target and performs multiple detections on the same target to obtain detection information of targets located at different distances.
  • the detection information is processed to determine the scattering intensity threshold range to improve the accuracy of the process of determining the credibility of the target.
  • S202 Determine the credibility of the type of the detection target according to the detection information of the detection target and the information of the Doppler units around the detection target.
  • the speed of the pedestrian target will not exceed a certain predetermined value, and the scattering intensity of the pedestrian target is weaker than that of the vehicle.
  • the reliability of the type of the detection target is determined according to the acquired detection information of the detection target. For example, by acquiring information such as the speed, scattering intensity, and distance of the detection target, the credibility of the type of the detection target can be determined.
  • the pedestrian target is taken as an example.
  • the acquired speed information of the detection target can be used to determine the reliability of the detection target type in combination with the speed ranges corresponding to different types of targets.
  • the credibility that the detection target is a pedestrian is added, for example, the first preset value is increased.
  • the reliability that the detection target is a pedestrian is deducted, for example, a second preset value is subtracted.
  • the acquired scattering intensity of the detection target may be compared with a pre-generated scattering intensity average and distance information change curve, or a pedestrian target's average scattering intensity and distance information look-up table.
  • the credibility that the detection target is a pedestrian is added, for example, the first preset value is increased.
  • the credibility that the detection target is a pedestrian is subtracted, for example, a second preset value is subtracted.
  • the detection target when the acquired speed information of the detection target meets the speed range corresponding to the pedestrian target, and when the acquired scattering intensity of the detection target meets the preset scattering intensity range corresponding to the pedestrian target, then the detection target is credible Points will be added to the degree; otherwise, points will be deducted for the reliability that the detection target is a pedestrian.
  • the movement of the target is more complicated, not only has overall movement, but its components also have micro movements such as acceleration, vibration, rotation, and roll.
  • the micro-motion of the target has corresponding micro-Doppler characteristics, which contains the landmark information of the target type such as movement and behavior, reflecting the fine characteristics of the target.
  • the micro-motions of different types of targets have different micro-Doppler characteristics, so that the target recognition is unique. For example, due to the swing of the limbs when the human body moves, it has obvious micro-Doppler characteristics, which contains landmark information such as human movement and behavior. Extract landmark information, such as Doppler energy, from the micro-Doppler features, thereby effectively identifying pedestrian targets.
  • the information of the Doppler unit around the detection target is acquired and analyzed, and then the corresponding micro-Doppler features are obtained, and landmark information, such as Doppler energy, is extracted therefrom, so as to obtain Doppler energy greater than
  • the Doppler unit with a preset Doppler energy threshold is thus determined according to the number of Doppler units with a Doppler energy greater than the preset Doppler energy threshold.
  • the credibility that the detection target type is a pedestrian is a pedestrian.
  • the reliability that the detection target is a pedestrian is added; and/or, When the number of Doppler units with Doppler energy greater than the preset Doppler energy threshold is less than the Doppler unit number threshold, the reliability that the detection target is a pedestrian is reduced.
  • the method for determining the reliability of the type of the detection target is specifically as follows:
  • the detection information includes scattering intensity information.
  • the scattering intensity is within the preset scattering intensity threshold range and the information of the Doppler unit around the detection target meets the preset Doppler unit condition, the credibility is increased by the first preset value; otherwise , The second preset value is subtracted from the credibility.
  • the detection information includes speed information.
  • the speed information is within the preset speed threshold range and the Doppler unit information around the detection target meets the preset Doppler unit condition, the reliability is increased by the first preset value; otherwise, The credibility is subtracted from a second preset value.
  • the detection information includes scattering intensity information and velocity information.
  • the velocity information is within the preset velocity threshold range, the scattering intensity and the preset scattering intensity threshold range, and the Doppler unit information around the detection target meets the preset Doppler unit condition.
  • the credibility is increased by the first preset value; otherwise, the credibility is reduced by the second preset value.
  • it further includes determining the type of the target according to the determined credibility of the type of the detection target.
  • at least one credibility range for different types of targets is preset. When the credibility of the determined type of the detection target is within a certain credibility range, the target is determined to be the target type corresponding to the credibility range.
  • the target credibility determination method provided by the embodiment of the present application obtains the detection information of the detection target and the information of the Doppler unit around the detection target, and according to the detection information of the detection target and the detection target The information of the surrounding Doppler units determines the credibility of the detection target type. Therefore, the credibility of the target type is determined based on the information obtained by the radar. The speed of system calculation is improved, which is conducive to the accuracy of determining the type of detection target, and is conducive to improving the safety and stability of the automatic driving/assisted driving process.
  • FIG. 3 is a schematic flowchart of a method for determining target credibility provided in an embodiment of the present application. See Figure 3, which specifically includes:
  • Step S301 Obtain the number of Doppler units whose scattering intensity is within a preset scattering intensity threshold range.
  • the energy value, distance value and speed value of a large number of scattered points (Doppler units) of the detection target can be obtained by radar.
  • the relationship between the energy value and the distance value of each Doppler unit can constitute a Doppler-distance plane, and the Doppler-distance plane includes a plurality of Doppler units.
  • the radar before acquiring the Doppler energy of each Doppler unit around the detection target of the detection target, it further includes preprocessing the echo signal obtained by the radar to obtain each Doppler energy around the detection target.
  • Doppler energy of the Doppler unit For example, the radar adopts FFT (Fast Fourier Transform, Fast Fourier Transform) technology to coherently accumulate the echo signals obtained by the radar to obtain the Doppler energy of each Doppler unit around the detection target.
  • FFT Fast Fourier Transform, Fast Fourier Transform
  • the preset Doppler energy threshold can be obtained by related methods. After the Doppler energy of each Doppler unit around the detection target is obtained by radar, the preset Doppler energy threshold can be used to detect the Doppler energy around the target.
  • the puller unit performs filtering, thereby filtering some invalid Doppler units, avoiding these invalid Doppler units from affecting subsequent judgments and reducing the amount of calculation.
  • the above-mentioned filtered Doppler unit may also be subjected to secondary filtering, for example, using the one-ring rule to output a higher quality Doppler unit, so as to further reduce the amount of calculation and improve processing efficiency.
  • the specific method for determining the preset Doppler energy threshold can be referred to but not limited to the following methods:
  • the radar can collect energy information and distance information of the target at different times in advance. For example, during the movement of the movable platform, the radar installed on the movable platform can continuously collect the Doppler energy information and distance information of the target. Then, according to the Doppler energy information and the distance information, the change curve of the radar energy and the distance is obtained. Finally, a preset Doppler energy threshold curve is determined according to the change curve.
  • the preset Doppler energy threshold can also be corrected according to the noise data of the radar. For example, the bottom noise of the radar can be added to the preset Doppler energy threshold to obtain the revised preset Doppler energy threshold.
  • other methods can also be used to modify the preset Doppler energy threshold, which is not limited.
  • Step S302 According to the number of Doppler units, determine whether the number of Doppler units is within a preset number of Doppler units, and determine the credibility.
  • a pedestrian target is taken as an example for description. Specifically: if the acquired Doppler energy of the detection target is greater than the preset Doppler energy threshold, the number of Doppler units is within the range of the preset Doppler unit number corresponding to the pedestrian target, then The reliability is increased by a third preset value; and/or, if the number of Doppler units is not within the range of the preset Doppler unit number corresponding to the pedestrian target, then the reliability is subtracted from the fourth preset value. Set value.
  • the specific method for determining the range of the preset Doppler unit number can be referred to the following method:
  • the radar obtains the number of Doppler units whose Doppler energy at different distances of the pedestrian target is greater than a preset Doppler energy threshold. After that, the distance information and the number of the Doppler units for which the Doppler energy of the pedestrian target corresponding to the distance information is greater than the preset Doppler energy threshold are recorded at the same time.
  • the obtained Doppler energy of pedestrian targets located at different distances is greater than the preset Doppler energy threshold and the number of Doppler units is averaged to form the threshold average value of the number of Doppler units and the distance information change A look-up table for the average value and distance information of the Doppler unit number threshold of the curve or pedestrian target.
  • the method for determining the credibility of the target obtained by the embodiment of the present application obtains the number of detection units with Doppler energy greater than the preset Doppler energy threshold among the Doppler units around the detection target, and determines the credibility of the target type It is beneficial to the accuracy of determining the type of detection target, and is beneficial to improving the safety and stability in the process of automatic driving/assisted driving.
  • FIG. 4 is a schematic flowchart of a method for determining target credibility provided in an embodiment of the present application, which specifically includes:
  • Step S401 Obtain the distance information
  • the radial distance of the detection target relative to the radar or the movable platform on which the radar is installed is obtained.
  • Step S402 Determine whether the distance information is less than a preset distance according to the distance information
  • the description will be given by taking a radar mounted on a car as an example.
  • the preset distance is determined in advance.
  • the fixed value of the preset distance is determined in advance, or the preset distance at different vehicle speeds is determined according to the speed of the car with the radar installed.
  • the determination of the preset distance is also related to at least one of the following factors: weather conditions, light intensity, driver's vision, braking equipment, and road conditions. The method of determining the preset distance based on any of the above factors is not limited here.
  • the preset distance is greater than or equal to the safe vehicle distance, that is, the preset distance is greater than or equal to the necessary separation distance between the vehicle equipped with the radar and the detection target during driving.
  • the preset distance is greater than or equal to the necessary separation distance between the vehicle equipped with the radar and the detection target during driving.
  • Step S403 If the distance information is less than the preset distance, control the movable platform to perform obstacle avoidance operations, or control the alarm device to perform alarm processing.
  • controlling the alarm device to perform alarm processing includes: according to the distance information, the movable platform installed with the radar controls the alarm device to display an alarm through LED lights, or through a digital display, or through voice broadcast, or through vibration.
  • the alarm device may be the movable platform, or other control platforms, etc., or may be the APP of the corresponding device. So as to ensure the safety in the process of automatic driving/assisted driving.
  • the target credibility determination method provided by the embodiment of the application obtains the distance information of the detected target and determines whether the distance information is less than the preset distance. Once the distance information is less than the preset distance, the movable platform is controlled to execute the avoidance Fault operation, or control the alarm device for alarm processing. In this way, passengers or drivers can know the abnormal situation in the driving process in time, which ensures the safety in the automatic driving/assisted driving process.
  • Fig. 5 is a schematic flowchart of a target recognition method provided by another embodiment of the present application, including:
  • Step S501 Detect the same detection target multiple times.
  • the detection target When detecting the detection target, it may be interfered during the detection process, which may cause excessive differences in the detection results during successive detection. As a result, the difference in the reliability of the detection results determined according to the successive detection times is too large, and the result of determining the target as a certain target type is problematic. For example, multipath interference may be encountered when the detection target is detected by radar, which may affect the detection result. The following is an example of detecting the detection target by radar.
  • the radar obtains the detection result of the same detection target once every time the radar detects the same detection target.
  • the radar can simultaneously or asynchronously detect multiple detection targets separately, and obtain the detection results of each detection target separately.
  • the detection result includes, but is not limited to, detection information of the detection target and information of Doppler units around the detection target.
  • the radar obtains the current detection result of the detection target according to the echo signal.
  • the current detection result of the detection target is obtained through the echo signal of the chirp continuous wave radar.
  • Step S502 Determine the current credibility of the detection target according to the current detection result of the detection target and the last credibility of the detection target.
  • the radar detects the same target each time, and can determine the credibility based on the detection information acquired each time.
  • the determination of the current credibility depends on the detection information acquired at the current moment and the last credibility. Therefore, as the number of detections performed by the radar for the same detection target increases, the current credibility of the detection target is constantly updated.
  • the current credibility of the detection target is determined by the following method: if the current detection result meets the second preset condition, the current credibility is the last available The credibility is increased by the first preset value; otherwise, the current credibility is the last credibility minus the second preset value.
  • the current detection result includes, but is not limited to: detection information of the detection target at the current moment and information of Doppler units around the detection target.
  • the second preset condition the detection information of the detection target at the current moment meets the preset detection information condition, and the information of the Doppler units around the detection target meets the preset Doppler condition.
  • the current credibility is the last time The credibility is increased by the first preset value; otherwise, the current credibility is the last credibility minus the second preset value.
  • the target credibility determination method involved in the embodiments of the present application can also be used in the foregoing target recognition method embodiments.
  • the "preset detection information condition" described in the foregoing embodiment corresponds to the "first preset condition” involved in an embodiment of the present application shown in FIG. 2.
  • the “preset Doppler condition” described in the above embodiment corresponds to the “preset Doppler unit number range” and the “preset Doppler unit number range” involved in an embodiment of the present application shown in FIG. Puller energy threshold".
  • multiple detection targets may be detected separately, and the detection results may be obtained separately.
  • the acquired multiple detection results of multiple detection targets can be processed simultaneously or asynchronously. In this way, it is possible to determine the target type for multiple detection targets, and to improve the safety and reliability of the operation of the automatic driving/assisted driving system.
  • Step S503 Determine that the detection target is a specific target type according to the current credibility.
  • the current credibility of the detection target represents the credibility that the detection target is a specific target type at the current moment. As the number of detections of the same detection target by radar increases, the current credibility of the detection target is constantly updated. Therefore, when the same target is detected multiple times to determine and update the current credibility of the detected target, the current credibility of the detected target can accurately determine whether the detected target is of a specific target type.
  • the type of the detection target is determined according to the number of detections of the same detection target and the current credibility. Specifically, it is described as an example to determine whether the detection target is a pedestrian target.
  • the radar detects the detection target for the first time, obtains the first detection result, and determines the first credibility that the detection target is a pedestrian target based on the detection result; subsequently, the radar detects the detection target for the second time and obtains the second detection result.
  • points are added or subtracted on the basis of the first credibility to determine the second credibility that the detection target is a pedestrian target. And so on to the Nth time.
  • the target recognition method provided by the embodiment of the present application detects the same detection target multiple times, and determines the current credibility of the detection target according to the current detection result of the detection target and the previous credibility of the detection target; And according to the current credibility, it is determined that the detection target is a specific target type. In this way, it is determined whether the target is a specific target type based on the information obtained by the radar. It effectively avoids the problem of determining the target type caused by abnormal detection results, improves the accuracy of target recognition, and then helps to improve the safety and stability of the automatic driving/assisted driving process.
  • FIG. 6 shows a schematic flowchart of a target recognition method provided by another embodiment of the present application, including:
  • Step S601 When the number of times of detecting the same detection target is less than the first preset number of detections, the current credibility of the detection target determined by each detection of the same detection target is compared with the first prediction. Set credibility for comparison.
  • a threshold value of the number of detection times for detecting the same detection target is preset as a stopping condition for multiple detections of the same detection target. For example, set the first preset detection times as the detection times threshold.
  • Set a credibility threshold as a judgment condition to determine whether the detection target is a pedestrian target.
  • the first preset credibility is set as the credibility threshold. As the number of detections performed by the radar for the detection target increases, the current credibility of the detected target is updated on the basis of the previous credibility.
  • the current credibility of the detection target and the first preset credibility determined each time the detection of the same detection target is determined Compare. Once the detection times of the detection target are greater than or equal to the first preset detection times, the detection target is no longer detected, so there is no need to detect the same detection target to determine the current credibility of the detection target Compare with the first preset credibility. Thereby, the calculation speed is improved, and the efficiency of determining that the detection target is a specific target type is improved.
  • Step S602 If the current credibility of the detection target that is detected and determined for the same detection target each time is less than the first preset credibility, it is determined that the detection target is not the specific target type And/or, if the current credibility is greater than the first preset credibility, it is determined that the detection target is the specific target type.
  • the first preset number of detections is used as the threshold of the number of detections for the same detection target, that is, as a stopping condition for multiple detections of the same detection target.
  • the number of times of detecting the same detection target multiple times reaches the first preset number of detections, stop detecting the same detection target. Therefore, within a limited number of times, the determination of whether the detection target is the specific target type is completed.
  • the reliability of each detection target determined by detecting the same detection target is less than the first preset Credibility, it is determined that the detection target is not a pedestrian target.
  • the current credibility of the detection target is compared with the first preset credibility For comparison, if the current credibility of the detection target is greater than the first preset credibility, it is determined that the detection target is a pedestrian target. The same detection target is no longer detected, and it is no longer determined whether the detection target is a pedestrian target.
  • the type of the detection target is determined according to the current credibility of the detection target.
  • the target is determined to be a specific category
  • a fixed value determination is made; if the target is not determined to be a specific category at the first preset number of times, it is determined that the target is not a specific target type.
  • FIG. 7 is a schematic flowchart of a target recognition method provided by another embodiment of the present application, including:
  • Step S701 When the number of times of detecting the same detection target is less than the first preset number of detections, and the current credibility is greater than the first preset credibility, the first preset credibility is reduced Small fifth preset value.
  • the radar When the radar detects the target, due to the phenomenon of multipath interference, the signal returned from the target reaches the radar antenna through different paths, which makes the radar work unstable and leads to errors in the radar detection results. By detecting the same detection target multiple times, the detection error can be effectively eliminated, thereby helping to improve the radar's ability to recognize target categories.
  • the detection target is a pedestrian target.
  • the target is determined to be a pedestrian target based on the detection result obtained at the previous detection moment; and the target is determined not to be a pedestrian target based on the detection result obtained at the current and future detection moments. If the number of detections performed on the same detection target is less than the first preset detection number, after the detection result obtained at the previous detection time is determined to be a pedestrian target, the detection target is no longer continued In detection, if the detection result obtained at the previous detection time is wrong, there may be a misjudgment to determine whether the detection target is a pedestrian target.
  • the target detection exit hysteresis judgment is set. For example, in the case that the number of times of detecting the same detection target is less than the first preset number of detections, each time the same detection target is detected, the current credibility is compared with the first preset credibility. If the current credibility is greater than the first preset credibility, the first preset credibility is updated, for example, the fifth preset value is decreased. Therefore, after detecting the same detection target next time, what needs to be compared is the next credibility and the updated first preset credibility.
  • the credibility of each detection target determined by detecting the same detection target is less than The first preset credibility determines that the detection target is not a pedestrian target.
  • the determination of whether the detection target is a pedestrian target is exited at this time, and there is no need to update the first preset credibility.
  • Step S702 After the number of times of detecting the same detection target is less than the first preset number of detections, and the current credibility is greater than the first preset credibility, perform the detection on the detection target The detection is continued, wherein the target type of the detection target is determined before the second preset number of detections is continued.
  • the current credibility is compared with the first preset probability. Reliability, if the current credibility is greater than the first preset credibility, continue detecting the detection target. This helps to avoid misjudgment of the type of detection target at the current moment due to incorrect detection results obtained at the previous detection moment.
  • a second number of detections is set as a termination condition for determining the target type of the detection target. In the process of continuing the detection of the detection target, the target type of the detection target needs to be determined before the number of continued detections reaches the second detection number. In this way, the type of detection target can be accurately determined in the limited detection process, processing time is reduced, and the safety of automatic driving/assisted driving is improved.
  • Step S703 If the current credibility of the detection target that is detected and determined on the same detection target each time is greater than or equal to a first preset credibility, it is determined that the detection target is the specific target type; And/or, if the current credibility of the detection target for detecting and determining the same detection target is less than a first preset credibility, it is determined that the detection target is not the specific target type.
  • a threshold value is preset, and the threshold value is used for Determine whether to determine whether the detection target is a specific target type.
  • the detection target when the detection target is continued to be detected, once the number of continued detections reaches the threshold, the detection target is determined to be a specific target type.
  • the detection target can be continued afterwards.
  • the threshold may also be used as a stopping condition for continuing detection of the same detection target, and once the number of continued detections reaches the threshold, the continuous detection of the same detection target is stopped. Therefore, the problem of large amount of system calculation and slow calculation speed caused by continuous acquisition and analysis of streaming data information is avoided.
  • the second preset number of detections is set as the threshold value.
  • the number of continued detections of the detection target is less than the second preset number of detections.
  • the alarm device when it is determined that the detection target is a specific target type, the alarm device is controlled to perform alarm processing.
  • the alarm device may be the movable platform or other control platforms. So as to ensure the safety in the process of automatic driving/assisted driving.
  • the target type of the detection target is output through LED light display, voice broadcast, and vibration.
  • the movable platform installed with the radar controls the alarm device to display an alarm through LED lights, or through a digital display, or through voice broadcast, or through vibration based on the detection target being a specific target type.
  • the radar is a millimeter wave radar.
  • the target recognition method provided by the embodiments of the present application sets the target detection exit lag judgment, which effectively solves the abnormal results caused by the detection interference, effectively improves the accuracy of target recognition, and is beneficial to improve automatic driving/assisted driving Safety and stability in the process.
  • the detection target is a pedestrian target.
  • FIG. 8A shows a type of Doppler-distance plane information provided by a specific embodiment of the present application, see FIG. 8A.
  • each frame of data acquisition corresponds to the acquisition of a frame of Doppler-distance unit plane image.
  • Pedestrian targets have extended Doppler due to the swinging movements of arms and legs when walking, so there are multiple Doppler units around the target.
  • the Doppler energy of multiple Doppler cells existing around the pedestrian target is low. According to the above characteristics, it can be determined whether the detection target is a pedestrian target.
  • Fig. 8B shows an algorithm implementation scheme of a specific embodiment of the present application. As shown in Figure 8B, the specific algorithm execution steps are as follows:
  • Multiple field tests are used to obtain the scattering intensity of pedestrian targets at different distances, and the scattering intensity is averaged, and then the scattering intensity threshold lookup table is obtained. And obtain the energy fluctuation range according to multiple measurements, such as ⁇ P_a, where P_a is a constant.
  • the scattering intensity threshold is selected according to the distance of the detection target. If the scattering intensity of the detection target is within the set threshold ⁇ P_a, the scattering intensity condition of the pedestrian target is satisfied.
  • the speed of setting the pedestrian target does not exceed 10m/s, and the speed range is adjustable. If the speed of the detected target exceeds the speed range, the target is considered not a pedestrian target.
  • the most appropriate Doppler energy threshold is selected through multiple trainings, and the Doppler energy threshold range is determined according to the intensity of the radar system noise signal. Specifically, the average Doppler energy of the non-target is counted, for example, the result is P_n, and the Doppler energy threshold + P_n is selected as the Doppler energy threshold range, where P_n is a constant.
  • the count is increased by 1, and when the count exceeds the threshold for the number of Doppler units, the Doppler of the pedestrian target is met Features, that is, satisfying Doppler judgment conditions.
  • the threshold of the number of Doppler units is obtained by taking the mean value of multiple pedestrian targets.
  • the radar detects the detection target once, obtains the detection data once, and determines the probability that the target is a pedestrian. That is, one frame of detection data is acquired for each detection, and the credibility that the target is a pedestrian is determined once.
  • the pedestrian probability Prob_i of the detection target increases with the probability Prob1; if the target does not meet the intensity decision, the pedestrian probability Prob_i decreases Go to the probability Prob2; if the target does not meet the speed decision, the pedestrian probability Prob_i is subtracted from the probability Prob3, where Prob1, Prob2, and Prob3 are constants and are obtained by training with multiple pedestrian targets.
  • a single frame of data is used to determine whether the detection target is a pedestrian target. Since the detection process may be interfered, a lot of misjudgments will occur. Therefore, the joint judgment is based on multi-frame accumulation, that is, through multiple detections of the same detection target, the current credibility of the detection target is determined according to the current detection result of the detection target and the previous credibility of the detection target; and according to the current credibility To determine whether the detection target is a pedestrian target.
  • the target is judged as a pedestrian target. If the number of frames reaches the set number of frames, when the current credibility of the detection target has not reached the first preset credibility threshold, the target is determined as a non-pedestrian target.
  • the current frame determines that the detection target is a pedestrian target
  • the next frame is determined to be a non-pedestrian target, and then it is detected as a pedestrian target.
  • the present invention sets the pedestrian detection exit hysteresis judgment. After the detection target is judged as the pedestrian target in the current frame, the first preset credibility is updated to a smaller value, and the judgment is continued on the next frame.
  • the first preset credibility is determined to be a non-pedestrian target; if within the second preset number of detections, the credibility of the detection target is determined to be greater than or equal to the updated first preset credibility, Then the detection target is determined as a pedestrian target.
  • the pedestrian detection state changes to the non-pedestrian detection state.
  • the threshold setting value is smaller than the threshold in a), that is, the first preset reliability is updated to a smaller value.
  • This specific embodiment shows the determination of whether the detection target is a pedestrian target. It should be understood that all detection targets with micro-Doppler characteristics are suitable for determining the target type of the detection target through the technical solution described in this embodiment. For example, according to the micro-Doppler characteristic of the propeller of the UAV, it can be determined whether the detection target is the UAV.
  • FIG. 9 is a schematic structural diagram of a target credibility determination system provided by an embodiment of the application .
  • the target credibility determination system 90 includes a processor 901.
  • the processors 901 are configured to execute the technical solutions of the embodiments of the foregoing method for determining target credibility.
  • the target credibility determination system 90 further includes: a memory 902 and a radar 903.
  • the memory is used to store program code
  • the processor 901 calls the program code, and when the program code is executed, is used to perform the following operations:
  • the reliability of the type of the detection target is determined.
  • the radar 903 and the processor 901 are provided separately.
  • the radar 903 includes a processor 901.
  • the target credibility determination system provided in this embodiment can execute the technical solutions of the foregoing embodiment of the target credibility determination method, and the execution mode and beneficial effects are similar, and will not be repeated here.
  • FIG. 10 is a schematic structural diagram of a target recognition system provided by an embodiment of this application, as shown in FIG. 10 ,
  • the target recognition system 100 includes: the target recognition system 100 includes: a processor 1001.
  • processors 1001 there are one or more processors 1001 that work together or individually, and the processors are configured to execute the technical solutions of the embodiments of the foregoing target recognition method.
  • the target recognition system 100 further includes: a memory 1002 and a radar 1003.
  • the memory 1002 is used to store program codes;
  • the processor 1001 calls the program code, and when the program code is executed, is used to perform the following operations:
  • the detection target is a specific target type.
  • the radar 1003 and the processor 1001 are provided separately.
  • the radar 1003 includes a processor 1001.
  • the target recognition system system provided in this embodiment can execute the technical solutions of the embodiments of the target recognition system method described above, and the execution mode and beneficial effects are similar, and will not be repeated here.
  • FIG. 11 is a schematic structural diagram of a radar provided by an embodiment of the application. As shown in FIG. 11, the radar 110 includes an antenna 1101 and a processor 1102.
  • the processor 1102 is in communication connection with the antenna, and is configured to execute the technical solutions of the foregoing target credibility determination method embodiment, and/or, to execute the technical solutions of the target recognition method embodiment.
  • FIG. 12 is a movable platform 120 provided by an embodiment of the application.
  • the movable platform 120 It includes: a body 1201, a power system 1202, and a system 90 for determining the target credibility of the technical solution.
  • the movable platform 120 may be any of the following: robots, unmanned aerial vehicles, unmanned vehicles, ordinary vehicles, VR glasses, and AR glasses.
  • the movable platform of the embodiment shown in FIG. 12 can be used to implement the technical solutions of the foregoing target credibility determination method embodiment, and its implementation principles and technical effects are similar, and will not be repeated here.
  • FIG. 13 is a movable platform 130 provided by an embodiment of the application.
  • the movable platform 130 Including: a fuselage 1301, a power system 1302, and the above-mentioned technical solution target recognition system 100.
  • the movable platform 130 may be any of the following: robots, unmanned aerial vehicles, unmanned vehicles, ordinary vehicles, VR glasses, and AR glasses.
  • the movable platform of the embodiment shown in FIG. 13 can be used to implement the technical solutions of the foregoing target recognition method embodiment, and its implementation principles and technical effects are similar, and will not be repeated here.
  • this embodiment also provides a computer-readable storage medium on which a computer program is stored, and the computer program is executed by a processor to implement the target credibility determination method and/or target recognition method described in the foregoing embodiment .
  • the disclosed device, system, and method may be implemented in other ways.
  • the device embodiments described above are merely illustrative.
  • the division of the units is only a logical function division, and there may be other divisions in actual implementation, for example, multiple units or components may be combined or It can be integrated into another system, or some features can be ignored or not implemented.
  • the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may be in electrical, mechanical or other forms.
  • the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.
  • the functional units in the various embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.
  • the above-mentioned integrated unit can be realized in the form of hardware, or in the form of hardware plus software functional unit.
  • the above-mentioned integrated unit implemented in the form of a software functional unit may be stored in a computer readable storage medium.
  • the above-mentioned software functional unit is stored in a storage medium and includes several instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (processor) execute the method described in the various embodiments of the present invention. Part of the steps.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk and other media that can store program code .

Landscapes

  • Engineering & Computer Science (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Physics & Mathematics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • General Physics & Mathematics (AREA)
  • Electromagnetism (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

一种目标可信度确定方法、目标识别方法、系统、可移动平台、存储介质。该目标可信度确定方法包括：获取检测目标的检测信息以及所述检测目标周围的多普勒单元的信息（S201）；根据所述检测目标的所述检测信息以及所述检测目标周围的多普勒单元的信息，确定所述检测目标的类型的可信度（S202）。该目标识别方法包括：多次探测同一检测目标；根据所述检测目标的当前探测结果以及所述检测目标的上一次可信度，确定所述检测目标的当前可信度；根据所述当前可信度，确定所述检测目标为特定的目标类型。该方法采用雷达对检测目标的可信度进行确定，并对检测目标是否为特定的目标类型进行确定，提高了对目标类型进行识别的准确性。

Description

一种目标可信度确定方法、一种目标识别方法、系统、车辆及存储介质
本专利文件披露的内容包含受版权保护的材料。该版权为版权所有人所有。版权所有人不反对任何人复制专利与商标局的官方记录和档案中所存在的该专利文件或该专利披露。
技术领域
本申请实施例涉及无人驾驶技术领域，尤其涉及一种目标可信度确定方法、目标识别方法、系统、雷达、可移动平台、存储介质。
背景技术
随着无人驾驶行业的发展，辅助驾驶和自动驾驶都成为当下的研究热点，而在辅助驾驶和自动驾驶领域，对目标的识别对于实现无人驾驶至关重要。例如，对于行人、车辆、路牌等不同类别目标进行有效识别后，从而根据具体情景进行减速避让、紧急停车、障碍物绕行、变道、自动按站停靠等功能。
传统技术中，主要通过视觉传感器采集目标的图像以对目标进行识别。但是，视觉传感器在对目标进行识别时，不具备全天时、全天候的特点。例如，在光线较弱，以及雨雪、雾霾天气，视觉传感器采集图像的效果不佳，将会明显降低视觉传感器对于目标识别的效果。此时，如果单纯依靠视觉传感器，将有可能出现大量自动驾驶失效的场景。相比于视觉传感器，毫米波雷达在工作时具有全天时、全天候等优点，但是相关技术还未能实现通过毫米波雷达对目标进行有效识别。
发明内容
本申请实施例提供了一种目标可信度确定方法、目标识别方法、系统、雷达、可移动平台、存储介质。
第一方面，本申请实施例提供了一种目标可信度确定方法，该方法包括：
获取检测目标的检测信息以及所述检测目标周围的多普勒单元 的信息；
根据所述检测目标的所述检测信息以及所述检测目标周围的多普勒单元的信息，确定所述检测目标的类型的可信度。
第二方面，本申请实施例提供了一种目标识别方法，该方法包括：
多次探测同一检测目标；
根据所述检测目标的当前探测结果以及所述检测目标的上一次可信度，确定所述检测目标的当前可信度；
根据所述当前可信度，确定所述检测目标为特定的目标类型。
第三方面，本申请实施例提供了一种目标可信度确定系统，该系统包括一个或多个处理器，共同地或单独地工作，所述处理器用于执行以下操作：
获取检测目标的检测信息以及所述检测目标周围的多普勒单元的信息；
根据所述检测目标的所述检测信息以及所述检测目标周围的多普勒单元的信息，确定所述检测目标的类型的可信度。
第四方面，本申请实施例提供了一种目标识别系统，该系统包括一个或多个处理器，共同地或单独地工作，所述处理器用于执行以下操作：
多次探测同一检测目标；
根据所述检测目标的当前探测结果以及所述检测目标的上一次可信度，确定所述检测目标的当前可信度；
根据所述当前可信度，确定所述检测目标为特定的目标类型。
第五方面，本申请实施例提供了一种雷达，该雷达包括：
天线，所述天线用于获取回波信号；
处理器，与所述天线通信连接，用于执行本申请的第一方面和/或第二方面任一项技术方案所述的方法。
第六方面，本申请实施例提供了一种可移动平台，该可移动平台包括：
机体；
动力系统，安装在所述机体，用于提供动力；
以及本申请的第五方面中任一项技术方案所述的雷达。
第七方面，本申请实施例提供了一种计算机可读存储介质，其特征在于，其上存储有计算机程序，所述计算机程序被处理器执行以实 现本申请第一方面的技术方案提供的目标可信度确定方法和/或本申请第二方面的技术方案提供的目标识别方法的步骤。
在本申请实施例中，通过对目标进行探测，获取目标的检测信息以及目标周围的多普勒单元的信息，根据获取的信息对目标进行识别。可以在全天时、全天候的条件下，对目标进行有效的识别，提升自动驾驶/辅助驾驶的稳定性和安全性。
附图说明
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1示出了本申请一实施例提供的一种应用场景示意图；
图2示出了本申请一实施例提供的一种目标可信度确定方法的流程示意图；
图3示出了本申请一实施例提供的一种目标可信度确定方法的流程示意图；
图4示出了本申请一实施例提供的一种目标可信度确定方法的流程示意图；
图5示出了本申请一实施例提供的一种目标识别方法的流程示意图；
图6示出了本申请另一实施例提供的一种目标识别方法的流程示意图；
图7示出了本申请又一实施例提供的一种目标识别方法的流程示意图；
图8A示出了本申请一具体实施例提供的一种多普勒-距离平面图像，图8B示出了本申请一具体实施例的算法实现方案；
图9示出了本申请一实施例的一种目标可信度确定系统的结构示意图；
图10示出了本申请一实施例提供的一种目标识别系统的结构示意图；
图11示出了本申请一实施例提供的一种雷达的结构示意图；
图12示出了本申请一实施例提供的一种可移动平台的结构示意图；
图13示出了本申请一实施例提供的一种可移动平台的结构示意图。
具体实施方式
为了提高自动驾驶/辅助驾驶系统的安全性，本申请实施例提出一种目标可信度确定方法、目标识别方法、系统、移动平台、存储介质。可移动平台上搭载的至少一个传感器：视觉系统、激光雷达、毫米波雷达、超声波雷达，通过利用上述至少一种自动驾驶/辅助驾驶所需的传感器对目标进行识别。例如，通过可移动平台上搭载的至少一个上述传感器对目标进行识别。该目标可以是人、动物、树木、车辆、路牌、栅栏、无人机等。该目标识别方法可以用于确定目标的类型、确定目标的航迹，规划可移动平台的航迹、可移动平台作业等场景。
其中，可信度为衡量检测目标的类型的可信程度的评价指标。通过对检测目标进行检测，获取回波信号中返回的检测目标的信息，并根据所获取的检测目标的信息，生成检测目标的类型的可信度。可信度越高，证明确定检测目标的类型的可信程度越高；可信度越低，证明确定检测目标的类型的可信程度越低。通过对多次探测确定的可信度进行多次判定，最终确定检测目标的类型。
在一些实施例中，通过多次探测同一检测目标，并根据检测目标的当前探测结果以及所述检测目标的上一次的探测结果，确定所述检测目标为特定的目标类型，例如行人目标、无人机目标等。
在一些实施例中，通过该目标识别方法对于目标进行识别后，还可以记录并预测目标的航迹，从而对目标进行跟踪。
在一些实施例中，通过该目标识别方法对于目标进行识别后，还能根据目标的当前航迹，规划可移动平台的运动轨迹。例如，进行减速避让、紧急停车、障碍物绕行、变道、自动按站停靠等。
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请的一 部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
图1提供了本申请实施例提供的一种应用场景举例。参见图1，该可移动平台为无人车10，该可移动平台包括车体101和雷达102。雷达102安装在车体101上。
雷达可以安装在可移动平台上，例如，机器人、无人飞行器、无人车、普通车辆、VR眼镜、AR眼镜等。以雷达安装在无人车为例，雷达可以集成在车辆的一处或多处位置，或者是安装于车辆上的装置，例如，车载设备等，对此不做限制。其中，雷达可以为毫米波雷达，也可以为其它类型的雷达传感器，对此不做限制。雷达至少包括天线，天线用于接收回波信号。
在本实施例中，雷达102随着无人车10的移动而移动，进行对待检测目标的探测，以获取用于目标可信度确定、目标识别的数据-获取检测目标的检测信息以及所述检测目标周围的多普勒单元的信息，并根据所述检测目标的所述检测信息以及所述检测目标周围的多普勒单元的信息，确定所述检测目标的类型的可信度；根据所述检测目标的当前探测结果以及所述检测目标的上一次可信度，确定所述检测目标的当前可信度；根据所述当前可信度，确定所述检测目标为特定的目标类型。
本实施例中用于目标可信度确定、目标识别的雷达102可为毫米微波雷达。
图2为本申请一实施例提供的一种目标可信度确定方法的流程示意图。如图2所示，该方法可以包括：
S201、获取检测目标的检测信息以及所述检测目标周围的多普勒单元的信息。
执行主体可以是雷达或可移动平台安装的处理器。下面以执行主体为雷达为例对目标可信度确定方法进行详细介绍。
在一些实施例中，检测目标的检测信息至少包括如下一种：距离、散射强度、速度信息、角度信息、观测能量。其中，距离信息包括检测目标相对于雷达的径向距离；散射强度信息包括检测目标在雷达波照射下所产生的散射回波的回波强度；速度信息包括由多普勒频移确定的检测目标相对于雷达的径向速度数据；角度信息为目标物体相较 于雷达所处的角度；观测能量即为回波信号的能量。
具体的，雷达通过发射电磁波并接收反馈的回波信号，以获取检测目标的检测信息。可选的，回波信号为调频连续波(frequency modulated continuous wave，FMCW)，例如，快速扫描波形、三角波或锯齿波。
在一些实施例中，检测目标周围的多普勒单元的信息至少包括如下一种：雷达对检测目标进行检测时获取的多普勒-距离平面图像上的检测目标周围的多普勒单元、检测目标周围的每个多普勒单元的多普勒能量、以及多普勒能量大于预设多普勒能量阈值的多普勒单元的数目。以此确定多普勒能量大于预设多普勒能量阈值的多普勒单元的数目。
具体的，获取检测目标周围的多普勒单元的信息，可以包括但不限于：通过雷达获取多普勒-距离平面图像，该多普勒-距离平面图像包括目标周围的多个多普勒单元；比较该多普勒单元的能量值与预设多普勒能量阈值，并记录多普勒能量大于预设多普勒能量阈值的多普勒单元的数目。可选的，将该多普勒能量大于预设多普勒能量阈值的多普勒单元添加到多普勒单元集合；否则，丢弃该多普勒单元。
其中，预设多普勒能量阈值可以为预先设定好的固定值，也可以是一参考多普勒单元的能量值。多普勒能量阈值可以由雷达的噪声信号强度、以及不同类别目标对应的多普勒能量确定。通过统计噪声的平均能量，选取噪声的平均能量与常数Pn相加，作为多普勒能量阈值。常数Pn为雷达采集目标的能量信息。其中，通过对不同类别目标进行多次采集、训练获取Pn。
在本实施例中，所述检测目标的所述检测信息包括如下至少一种：距离、散射强度、速度信息；以及检测目标周围的多普勒单元的信息包括如下至少一种：检测目标周围的每个多普勒单元的多普勒能量、以及多普勒能量大于预设多普勒能量阈值的多普勒单元的数目为例进行说明。
其中，获取检测目标的检测信息以及所述检测目标周围的多普勒单元的信息，具体包括：
在获取所述检测目标的检测信息的同时，获取所述检测目标周围的多普勒单元的信息。即在对检测目标进行探测的同时或之后，就获取检测目标的检测信息、以及检测目标周围的多普勒单元的信息。
或者，先获取所述检测目标的检测信息，若所述检测目标的检测信息满足第一预设条件，则获取所述检测目标周围的多普勒单元的信息；和/或，若所述检测目标的检测信息不满足所述第一预设条件，则不再获取所述检测目标周围的多普勒单元的信息。对于检测信息先做一次判断，在满足第一预设条件的情况下，才进行获取检测目标周围的多普勒单元的信息。从而减少了获取检测目标周围的多普勒单元的信息的次数，减少了运算量，提高系统的运算速度。
具体的，判断检测信息是否满足第一预设条件具体包括但不限于如下方法：判断检测信息所包含的具体参数是否在其对应的阈值范围内。例如，判断散射强度是否在预设散射强度阈值范围内，所述速度信息是否在预设速度阈值范围内等。下面结合以下三种情形进行示例性说明：
例如，所述检测信息包括散射强度信息，若所述散射强度在预设散射强度阈值范围内，则所述检测目标的检测信息满足第一预设条件，进行获取检测目标周围的多普勒单元的信息。在一些实施例中，若所述散射强度不在预设散射强度阈值范围内，则所述检测目标的检测信息不满足所述第一预设条件，不进行获取检测目标周围的多普勒单元的信息。
又例如，所述检测信息包括速度信息，若所述速度信息在预设速度阈值范围内，则所述检测目标的检测信息满足第一预设条件，进行获取检测目标周围的多普勒单元的信息。在一些实施例中，若所述速度信息不在预设速度阈值范围内，则所述检测目标的检测信息不满足所述第一预设条件，不进行获取检测目标周围的多普勒单元的信息。
再例如，所述检测信息包括散射强度信息、速度信息，若所述散射强度在预设散射强度阈值范围内，所述速度信息在预设速度阈值范围内，则所述检测目标的检测信息满足第一预设条件，进行获取检测目标周围的多普勒单元的信息。即，只有同时满足两者的情况下，检测目标的检测信息才满足第一预设条件，才进行获取检测目标周围的多普勒单元的信息。在一些实施例中，若所述散射强度不在预设散射强度阈值范围内，所述速度信息不在预设速度阈值范围内，所述检测目标的检测信息不满足所述第一预设条件，不进行获取检测目标周围的多普勒单元的信息。
在上述实施例的基础上，确定所述散射强度阈值范围，具体包括 但不限于以下方法：
(1)、获取外部目标在不同距离处的所述散射强度。
同一种类别的目标，与雷达间隔的距离不同，雷达获取的回波信号中包含的检测信息也不相同。例如，在与雷达间隔的距离不同的情况下，雷达对于同一目标进行多次探测，以获取位于不同距离处的目标的散射强度。在一些实施例中，预先对获取到的位于不同距离处的目标的散射强度取平均值，以提高确定目标可信度过程的准确性。在一些实施例中，对于多种不同类别的目标，雷达分别对于其中的目标进行多次探测，以获取位于不同距离处的目标的散射强度。
本实施例中，以获取行人目标在不同距离处的散射强度为例。在外场试验中，雷达通过对行人目标进行检测，获取行人目标在不同距离处的散射强度。之后，同时记下距离信息以及该距离信息对应的散射强度信息。对获取到的位于不同距离处的行人目标的散射强度取平均值，形成散射强度平均值和距离信息变化曲线、或者行人目标的散射强度平均值和距离信息查找表。
(2)、根据所述获取外部目标在所述不同距离处的所述散射强度，确定所述散射强度阈值范围。
在一些实施例中，根据散射强度平均值和距离信息变化曲线确定散射强度阈值。可选的，考虑到不同类型的目标具有不同的散射强度，因此对于不同类型的目标确定各自的散射强度阈值范围。在一些实施例中，考虑到雷达具有噪声，因此散射强度阈值还由雷达的噪声信号强度确定。通过统计噪声的平均强度，选取噪声的平均强度与获取到的目标在不同距离处的散射强度相加，作为散射强度阈值。
本实施例中，在可移动平台的运动或静止的过程中，安装在移动平台的雷达探测到外部目标，对于同一目标进行多次探测，以获取位于不同距离处的目标的检测信息，并对检测信息进行处理，确定所述散射强度阈值范围，以提高确定目标可信度过程的准确性。
S202、根据所述检测目标的所述检测信息以及所述检测目标周围的多普勒单元的信息，确定所述检测目标的类型的可信度。
一方面，不同类别的目标，具有不同的运动速度和/或散射强度。例如，行人目标的速度不会超过某预定值，行人目标的散射强度相对于车辆较弱。在一些实施例中，根据获取到的检测目标的检测信息，确定检测目标的类型的可信度。例如，通过获取检测目标的速度、散 射强度、距离等信息，确定检测目标的类型的可信度。
具体的，以行人目标为例进行说明。例如，可以利用获取到的检测目标的速度信息，结合不同类型的目标对应的速度范围对检测目标的类型的可信度进行确定。当获取到的检测目标的速度信息满足行人目标对应的速度范围，则对检测目标为行人的可信度进行加分，例如增加第一预设值。和/或，当获取到的检测目标的速度信息不满足行人目标对应的速度范围，则对检测目标为行人的可信度进行减分，例如减去第二预设值。
又例如，可以将获取到的检测目标的散射强度，与预先生成的形成散射强度平均值和距离信息变化曲线、或者行人目标的散射强度平均值和距离信息查找表进行比较。当获取到的检测目标的散射强度满足行人目标对应的预设散射强度范围，则对检测目标为行人的可信度进行加分，例如增加第一预设值。和/或，当获取到的检测目标的散射强度不满足行人目标对应的预设散射强度范围，则对检测目标为行人的可信度进行减分，例如减去第二预设值。
再例如，当获取到的检测目标的速度信息满足行人目标对应的速度范围，以及当获取到的检测目标的散射强度满足行人目标对应的预设散射强度范围，则对检测目标为行人的可信度进行加分；否则对检测目标为行人的可信度进行减分。
另一方面，目标的运动情况较为复杂，不仅具有整体的运动，其部件还具有加速、震动、旋转、翻滚等微运动。目标的微运动具有相应的微多普勒特征，其中含有的与运动、行为等目标类型的标志性信息，反映了目标的精细特征。不同类别的目标的微运动具有不同的微多普勒特征，从而使目标的识别具有唯一性。例如，人体运动时由于肢体的摆动，具有明显的微多普勒特征，其中含有与人体运动、行为等标志性信息。从微多普勒特征中提取标志性信息，如多普勒能量，从而有效地识别行人目标。
在一些实施例中，获取并分析检测目标周围的多普勒单元的信息，进而得到相应的微多普勒特征，并从中提取标志性信息，如多普勒能量，进而获得多普勒能量大于预设多普勒能量阈值的所述多普勒单元，从而根据多普勒能量大于预设多普勒能量阈值的所述多普勒单元的数目确定所述可信度。通过获取并分析检测目标周围的多普勒单元的信息，对确定目标类型的可信度具有更高的准确性。
具体的，以确定所述检测目标的类型为行人的可信度为例进行说明。当多普勒能量大于预设多普勒能量阈值的所述多普勒单元的数目大于等于多普勒单元数目阈值时，则对检测目标为行人的可信度进行加分；和/或，当多普勒能量大于预设多普勒能量阈值的所述多普勒单元的数目小于多普勒单元数目阈值时，则对检测目标为行人的可信度进行减分。
在本实施例中，根据所述检测目标的所述检测信息以及所述检测目标周围的多普勒单元的信息，确定所述检测目标的类型的可信度的方法具体如下：
所述检测信息包括散射强度信息。当所述散射强度在预设散射强度阈值范围内，且所述检测目标周围的多普勒单元的信息满足预设多普勒单元条件时，所述可信度增加第一预设值；否则，所述可信度减去第二预设值。
或者，所述检测信息包括速度信息。当所述速度信息在预设速度阈值范围内，且所述检测目标周围的多普勒单元的信息满足预设多普勒单元条件时，所述可信度增加第一预设值；否则，所述可信度减去第二预设值。
或者，所述检测信息包括散射强度信息、速度信息。当所述速度信息在预设速度阈值范围内，所述散射强度与预设散射强度阈值范围内，且所述检测目标周围的多普勒单元的信息满足预设多普勒单元条件时，所述可信度增加第一预设值；否则，所述可信度减去第二预设值。
在一些实施例中，还包括根据所确定的所述检测目标的类型的可信度，确定目标的类型。可选的，预先设定不同类型的目标的至少一个可信度范围。当所确定的所述检测目标的类型的可信度在某一可信度范围，则确定目标为该可信度范围对应的目标类型。
本申请实施例提供的目标可信度确定方法，通过获取检测目标的检测信息以及所述检测目标周围的多普勒单元的信息，并根据所述检测目标的所述检测信息以及所述检测目标周围的多普勒单元的信息，确定所述检测目标的类型的可信度。从而基于雷达获取的信息确定目标类型的可信度。提高了系统运算的速度，有利于对于检测目标的类型进行确定的准确性，有利于提高自动驾驶/辅助驾驶过程中的安全性和稳定性。
在图2的基础上，图3是本申请一实施例中提供的一种目标可信度确定方法的流程示意图。请参见图3，具体包括：
步骤S301、获取所述散射强度在预设散射强度阈值范围内的所述多普勒单元的数目。
在本实施例中，具体包括：
(1)获取所述检测目标周围的每个多普勒单元的多普勒能量。
通过雷达可以获取检测目标的大量散射点(多普勒单元)的能量值、距离值和速度值等。可选的，每个多普勒单元的能量值和距离值的关系，就可以构成多普勒-距离平面，该多普勒-距离平面包括多个多普勒单元。
在一些实施例中，在获取检测目标的检测目标周围的每个多普勒单元的多普勒能量之前，还包括对雷达获取到的回波信号进行预处理，以获取检测目标周围的每个多普勒单元的多普勒能量。例如，雷达采用FFT(Fast Fourier Transform，快速傅里叶变换)技术对于雷达获取到的回波信号做相参积累，以获取检测目标周围的每个多普勒单元的多普勒能量。当然，在实际应用中，还可以采用其它技术获取，对此不做限制。
(2)根据所述每个多普勒单元的所述多普勒能量，确定所述散射强度在预设散射强度阈值范围内的所述多普勒单元的数目。
对于检测目标周围的每个多普勒单元进行能量筛选，例如，通过RCS(Radar Cross Section，雷达截面积)、或CFAR(Constant False-Alarm Rate，恒虚警)算法过滤。可以通过相关方法得到预设多普勒能量阈值，在通过雷达获取到检测目标周围的每个多普勒单元的多普勒能量后，可以利用预设多普勒能量阈值对检测目标周围的多普勒单元进行过滤，从而过滤一些无效的多普勒单元，避免这些无效的多普勒单元影响后续判断，并减少运算量。在一些实施例中，还可以对上述过滤的多普勒单元进行二次过滤，例如利用one-ring法则，以输出较高质量的多普勒单元，确定进一步减少运算量，提高处理效率。
其中，确定预设多普勒能量阈值的具体方法可以参见但不限于如下方法：
首先，对于不同类别的目标，雷达可以预先采集目标在不同时刻的能量信息和距离信息。例如，在可移动平台的运动过程中，安装在可移动平台的雷达可以不断采集目标的多普勒能量信息和距离信息。 然后，根据该多普勒能量信息和该距离信息获取雷达的能量与距离的变化曲线。最后，根据该变化曲线确定预设多普勒能量阈值曲线。
在一些实施例中，在一些实施例中，考虑到雷达具有噪声，因此还可以根据雷达的噪声数据修正预设多普勒能量阈值。例如，可以将雷达的底噪与预设多普勒能量阈值进行相加，得到修正后的预设多普勒能量阈值。当然，在实际应用中，还可以采用其它方式修正预设多普勒能量阈值，对此不做限制。
步骤S302、根据所述多普勒单元的数目，确定所述多普勒单元的数目是否在预设多普勒单元数目范围内，并确定所述可信度。
在本实施例中，以行人目标为例进行说明。具体包括：若获取到的检测目标的多普勒能量大于预设多普勒能量阈值的所述多普勒单元的数目在行人目标对应的预设多普勒单元数目范围内，则所述可信度增加第三预设值；和/或，若所述多普勒单元的数目不在行人目标对应的所述预设多普勒单元数目范围内，则所述可信度减去第四预设值。
其中，确定预设多普勒单元数目范围的具体方法可以参见如下方法：
以获取行人目标在不同距离处的、多普勒能量大于预设多普勒能量阈值的所述多普勒单元的数目为例进行说明。雷达通过对行人目标进行检测，获取行人目标在不同距离处的多普勒能量大于预设多普勒能量阈值的所述多普勒单元的数目。之后，同时记下距离信息以及该距离信息对应的行人目标的多普勒能量大于预设多普勒能量阈值的所述多普勒单元的数目。对获取到的位于不同距离处的行人目标的多普勒能量大于预设多普勒能量阈值的所述多普勒单元的数目取平均值，形成多普勒单元数目阈值平均值和距离信息变化曲线、或者行人目标的多普勒单元数目阈值平均值和距离信息查找表。
本申请实施例提供的目标可信度确定方法，通过获取检测目标周围的多普勒单元中，多普勒能量大于预设多普勒能量阈值的检测单元的数目，并确定目标类型的可信度，有利于对于检测目标的类型进行确定的准确性，有利于提高自动驾驶/辅助驾驶过程中的安全性和稳定性。
在图2所示实施例的基础上，图4是本申请一实施例中提供的一种目标可信度确定方法的流程示意图，具体包括：
步骤S401、获取所述距离信息；
具体的，获取检测目标相对于雷达或安装雷达的可移动平台的径向距离。
步骤S402、根据所述距离信息，判断所述距离信息是否小于预设距离；
具体的，以雷达搭载在汽车上为例进行说明。根据获取的检测目标相对于雷达或安装雷达的汽车的径向距离，判断该距离信息是否小于预设距离。可选的，在多次外场试验中，预先确定好预设距离。例如，预先确定好预设距离的固定值，或者，根据安装雷达的汽车车速，确定不同车速下的预设距离。可以理解的是，预设距离的确定还与如下至少一个因素相关：天气情况、光照强度、司机视力、刹车设备、路面状况。根据上述任意因素确定预设距离的方法，在此不作限定。
其中，预设距离大于等于安全车距，即预设距离大于等于搭载雷达的汽车在行驶中与检测目标所保持的必要间隔距离。以便在遇到紧急情况时留有足够的刹车空间(包括刹车时间、刹车距离等)，有利于汽车的安全行驶。
步骤S403、若所述距离信息小于预设距离，则控制所述可移动平台执行避障操作，或者控制报警装置进行报警处理。
具体的，当所述距离信息小于预设距离，此时雷达、或者安装雷达的可移动平台距离检测目标的距离相对较近，为了保证自动驾驶/辅助驾驶过程中的安全性，可以控制所述可移动平台执行避障操作，和/或，控制报警装置进行报警处理。在一些实施例中，控制报警装置进行报警处理，包括：安装雷达的可移动平台根据所述距离信息，控制报警装置通过LED灯显示报警或者通过数字显示报警或者通过语音播报报警或者通过震动报警。所述报警装置可以是该可移动平台、或者其他控制平台等，也可以是对应设备的APP。从而保证自动驾驶/辅助驾驶过程中的安全性。
本申请实施例提供的目标可信度确定方法，通过获取检测目标的距离信息，并判断该距离信息是否小于预设距离，一旦该距离信息小于预设距离，则控制所述可移动平台执行避障操作，或者控制报警装置进行报警处理。从而乘客或驾驶员能够及时得知行驶过程中的异常情况，保证了自动驾驶/辅助驾驶过程中的安全性。
图5是本申请另一实施例提供的一种目标识别方法的流程示意 图，包括：
步骤S501、多次探测同一检测目标。
对检测目标进行探测时，可能在探测过程中会受到干扰因此造成先后探测时探测结果的差异过大。从而造成根据先后探测时刻确定的探测结果所分别确定的可信度差异过大，进而造成确定目标为确定的目标类型的结果出现问题。例如，通过雷达对检测目标进行探测时可能受到多径干扰，进而影响探测结果。下面以通过雷达对检测目标进行探测为例进行说明。
在一些实施例中，雷达每对所述同一检测目标进行探测一次，获取一次所述同一检测目标的所述探测结果。可选的，雷达可以同时或异步地对多个检测目标分别进行探测，分别获取每个检测目标的探测结果。可选的，探测结果包括但不限于检测目标的检测信息以及所述检测目标周围的多普勒单元的信息。
在一些实施例中，雷达根据回波信号获取所述检测目标的当前探测结果。可选的，通过线性调频连续波雷达回波信号获取所述检测目标的当前探测结果。
步骤S502、根据所述检测目标的当前探测结果以及所述检测目标的上一次可信度，确定所述检测目标的当前可信度。
雷达每次对于同一检测目标进行探测，能根据每次获取到的探测信息，确定一次可信度。雷达在当前时刻对于同一检测目标进行探测时，当前可信度的确定依赖于当前时刻获取到的探测信息以及上一次可信度。从而随着雷达对于同一检测目标进行探测的探测次数的增加，检测目标的当前可信度也不断得到更新。
在一些实施例中，针对当前的探测结果，通过下述方法确定检测目标的当前可信度：若所述当前探测结果满足第二预设条件，所述当前可信度为所述上一次可信度增加第一预设值；否则，所述当前可信度为所述上一次可信度减去第二预设值。
其中，当前探测结果包括但不限于：当前时刻检测目标的检测信息以及所述检测目标周围的多普勒单元的信息。可选的，第二预设条件：当前时刻检测目标的检测信息满足预设检测信息条件以及所述检测目标周围的多普勒单元的信息满足预设多普勒条件。
具体的，若当前时刻检测目标的检测信息满足预设检测信息条件以及所述检测目标周围的多普勒单元的信息满足预设多普勒条件时， 所述当前可信度为所述上一次可信度增加第一预设值；否则，所述当前可信度为所述上一次可信度减去第二预设值。
需要注意的是，本申请的实施例中所涉及到的目标可信度确定方法也可以用于上述目标识别方法的实施例。可选的，上述实施例所述的“预设检测信息条件”对应于图2所示的本申请的一个实施例所涉及到的“第一预设条件”。可选的，上述实施例所述的“预设多普勒条件”对应于图2所示的本申请的一个实施例所涉及到的“预设多普勒单元数目范围”以及“预设多普勒能量阈值”。
在一些实施例中，可以对多个检测目标分别进行探测，并分别获取探测结果。所获取到的多个检测目标的多个探测结果可以被同时处理，也可以异步处理。从而实现对多个检测目标进行目标类型的确定，提高自动驾驶/辅助驾驶系统运行的的安全性和可靠性。
步骤S503、根据所述当前可信度，确定所述检测目标为特定的目标类型。
检测目标的当前可信度表征检测目标在当前时刻为特定的目标类型的可信程度。随着雷达对于同一检测目标进行探测的探测次数的增加，检测目标的当前可信度也不断得到更新。因此多次探测同一目标，以确定并更新检测目标的当前可信度的情况下，通过检测目标的当前可信度能够准确的确定所述检测目标是否为特定的目标类型。
可选地，根据对所述同一检测目标进行探测的次数，以及所述当前可信度确定所述检测目标的类型。具体的，以确定检测目标是否为行人目标为例进行说明。雷达第一次探测检测目标，获取第一次的探测结果，并根据该检测结果确定检测目标是行人目标的第一次可信度；随后，雷达第二次探测检测目标，获取第二次的探测结果，根据该检测结果，在第一次可信度的基础上进行加分或者减分，确定检测目标是行人目标的第二次可信度。以此类推到第N次。最后根据第N次可信度确定检测目标是否为行人目标。
本申请实施例提供的目标识别方法，通过多次探测同一检测目标，根据所述检测目标的当前探测结果以及所述检测目标的上一次可信度，确定所述检测目标的当前可信度；并根据所述当前可信度，确定所述检测目标为特定的目标类型。从而基于雷达获取的信息确定目标是否为特定的目标类型。有效避免了因探测结果异常造成的目标类型的确定出现的问题，提高了对于目标识别的准确性，进而有利于提高 自动驾驶/辅助驾驶过程中的安全性和稳定性。
在图5所示实施例的基础上，图6示出了本申请又一实施例提供的一种目标识别方法的流程示意图，包括：
步骤S601、当所述对所述同一检测目标进行探测的次数小于第一预设探测次数，将每次对所述同一检测目标进行探测确定的所述检测目标的当前可信度与第一预设可信度进行比较。
具体的，以确定检测目标为行人目标为例进行说明。预先设置一个对所述同一检测目标进行探测的探测次数门限值，作为对所述同一检测目标进行多次探测的停止条件。例如设置第一预设探测次数作为探测次数门限值。设定一个可信度门限作为确定检测目标是否为行人目标的判断条件。例如设定第一预设可信度作为可信度门限。随着雷达对于检测目标进行探测的探测次数的增加，检测目标的当前可信度在上一次可信度的基础上得到更新。当雷达对于检测目标进行探测的探测次数小于第一预设探测次数时，将每次对所述同一检测目标进行探测确定的所述检测目标的当前可信度与第一预设可信度进行比较。一旦对检测目标进行探测的探测次数大于等于第一预设探测次数，就不再对检测目标进行探测，因此也无需再对所述同一检测目标进行探测确定的所述检测目标的当前可信度与第一预设可信度进行比较。从而提高了运算速度，提高了确定检测目标是特定的目标类型的效率。
步骤S602、若每次对所述同一检测目标进行探测确定的所述检测目标的当前可信度均小于所述第一预设可信度，则确定所述检测目标不是所述特定的目标类型；和/或，若一旦所述当前可信度大于所述第一预设可信度，则确定所述检测目标是所述特定的目标类型。
具体的，以确定检测目标为行人目标为例进行说明。可选的，使用第一预设探测次数作为对所述同一检测目标进行探测的探测次数门限值，也即作为对所述同一检测目标进行多次探测的停止条件。当对所述同一检测目标进行多次探测的探测次数达到第一预设探测次数时，停止对所述同一检测目标进行探测。从而在有限的次数内，完成对所述检测目标是否为所述特定的目标类型的确定。
例如，若直到对所述同一检测目标进行探测的探测次数达到第一预设探测次数时，对所述同一检测目标进行探测所确定的检测目标的每一次的可信度均小于第一预设可信度，则确定检测目标不是行人目标。从而减小了系统的运算量，提高对检测目标进行识别的效率。
又例如，当所述对所述同一检测目标进行探测的次数未达到第一预设探测次数时，每次对检测目标探测之后，对检测目标的当前可信度与第一预设可信度进行比较，若检测目标的当前可信度大于第一预设可信度，则确定检测目标为行人目标。并不再对该同一检测目标进行探测，并不再确定检测目标是否为行人目标。从而减小了系统的运算量，实现对检测目标的的快速识别。
本申请实施例提供的目标识别方法，在所述对所述同一检测目标进行探测的次数未达到第一预设探测次数时，根据检测目标的当前可信度对检测目标的类别进行确定。当确定目标为特定的类别后，定值判定；若在第一预设次数时还未确定出目标为特定的类别，则判定目标不是特定的目标类型。通过这种方法，有效提高了目标识别的效率，同时也提高自动驾驶/辅助驾驶过程中的安全性。
在图5所示实施例的基础上，图7是本申请又一实施例提供的一种目标识别方法的流程示意图，包括：
步骤S701、当对所述同一检测目标进行探测的次数小于所述第一预设探测次数，所述当前可信度大于所述第一预设可信度，则第一预设可信度减小第五预设值。
雷达在对目标进行探测时，由于存在多径干扰现象，从目标返回的信号通过不同路径到达雷达天线，使得雷达工作不稳定，导致雷达的探测结果存在误差。通过对同一检测目标进行多次探测，能够有效消除探测误差，从而有助于提高雷达对目标类别的识别能力。
具体的，以确定检测目标是行人目标为例进行说明。
在一些实施例中，根据前一个探测时刻所获取的探测结果，确定目标是行人目标；而根据当前及以后的探测时刻所获取的探测结果，确定目标不是行人目标。如果在对所述同一检测目标进行探测的次数小于所述第一预设探测次数的情况下，根据前一个探测时刻所获取的探测结果，确定目标是行人目标后，就不再对检测目标继续探测，前一个探测时刻所获取的探测结果有误的情况下，可能会出现对于确定检测目标是否是行人目标的误判断。
为避免出现上述情况，在一些实施例中，设定目标检测退出迟滞判断。例如，在对所述同一检测目标进行探测的次数小于所述第一预设探测次数的情况下，每次对所述同一检测目标进行探测后，比较当前可信度与第一预设可信度，若当前可信度大于所述第一预设可信度， 则对第一预设可信度进行更新，例如，减小第五预设值。从而，下一次对所述同一检测目标进行探测后，需要比较的是下一次可信度与更新后的第一预设可信度。
在一些实施例中，若直到对所述同一检测目标进行探测的探测次数达到第一预设探测次数时，对所述同一检测目标进行探测所确定的检测目标的每一次的可信度均小于第一预设可信度，则确定检测目标不是行人目标。可选的，此时退出对检测目标是否是行人目标的判定，也无需再对第一预设可信度进行更新。
步骤S702、当所述对所述同一检测目标进行探测的次数小于所述第一预设探测次数，所述当前可信度大于所述第一预设可信度后，对所述检测目标进行继续探测，其中，在继续探测第二预设探测次数之前对所述检测目标的目标类型进行确定。
具体的，在对所述同一检测目标进行探测的次数小于所述第一预设探测次数的情况下，每次对所述同一检测目标进行探测后，比较当前可信度与第一预设可信度，若当前可信度大于所述第一预设可信度，则对所述检测目标进行继续探测。从而有助于避免因前一个探测时刻所获取的探测结果有误而出现的对于当前时刻确定检测目标的类型的误判断。可选的，对所述检测目标进行继续探测之前，设置第二探测次数作为对所述检测目标的目标类型进行确定的终止条件。在对所述检测目标进行继续探测的过程中，在继续探测次数达到第二探测次数之前，需要对所述检测目标的目标类型进行确定。从而在有限次探测过程中准确的确定检测目标的类型，减少处理时间，提高自动驾驶/辅助驾驶的安全性。
步骤S703、若每次对所述同一检测目标进行探测确定的所述检测目标的当前可信度均大于等于第一预设可信度，则确定所述检测目标是所述特定的目标类型；和/或，对所述同一检测目标进行探测确定的所述检测目标的当前可信度均小于第一预设可信度，则确定所述检测目标不是所述特定的目标类型。
具体的，在对所述同一检测目标进行探测的次数小于所述第一预设探测次数的情况下，对所述检测目标进行继续探测之前，预先设置一个门限值，该门限值用于判断是否进行确定检测目标为特定的目标类型。
例如，当对所述检测目标进行继续探测时，若继续探测的次数一 旦达到该门限值，就对检测目标为特定的目标类型进行确定。可选的，之后可以继续对检测目标进行探测。可选的，该门限值还可以作为对所述同一检测目标进行继续探测的停止条件，继续探测的次数一旦达到该门限值，就停止对所述同一检测目标进行继续探测。从而避免因持续获取并分析流式数据信息而造成的系统运算量大、运算速度慢的问题。
在本实施例中，通过设置第二预设探测次数作为该门限值。对所述检测目标进行继续探测的次数小于第二预设探测次数。在此过程中，若每次对所述同一检测目标进行探测确定的所述检测目标的当前可信度均大于等于第一预设可信度，则确定所述检测目标为特定的目标类型。和/或，若每次对所述同一检测目标进行探测确定的所述检测目标的当前可信度均小于第一预设可信度，则确定所述检测目标不是特定的目标类型。通过这种方法，有效避免了因探测结果的抖动对目标类型的确定造成的误判。
在一些实施例中，当确定所述检测目标为特定的目标类型，则控制报警装置进行报警处理。具体的，所述报警装置可以是该可移动平台、或者其他控制平台等。从而保证自动驾驶/辅助驾驶过程中的安全性。
在一些实施例中，确定所述检测目标为特定的目标类型后，通过LED灯显示、语音播报、震动方式，输出所述检测目标的目标类型。具体的，安装雷达的可移动平台根据所述检测目标为特定的目标类型，控制报警装置通过LED灯显示报警或者通过数字显示报警或者通过语音播报报警或者通过震动报警。
在一些实施例中，所述雷达为毫米波雷达。
本申请实施例提供的目标识别方法，设定目标检测退出迟滞判断，有效解决了因探测受到干扰而产生的异常结果，有效提高了对于目标识别的准确性，进而有利于提高自动驾驶/辅助驾驶过程中的安全性和稳定性。
在一具体实施例中，以确定检测目标为行人目标为例进行说明。
图8A示出了本申请一具体实施例提供的一种多普勒-距离平面信息，参见图8A。其中，每获取一帧数据对应着获取一帧多普勒-距离单元平面图像，颜色越深，表征多普勒单元的多普勒能量越高；颜色越浅，表征多普勒单元的多普勒能量越低。行人目标因走路时存在 手臂和腿部的摆动动作，在多普勒上具有拓展特性，因此目标周围存在有多个多普勒单元。而行人目标周围存在的多个多普勒单元的多普勒能量较低。根据上述特征可与对检测目标是否是行人目标进行确定。
图8B示出了本申请一具体实施例的算法实现方案。如图8B所示，具体算法执行步骤如下：
(1)对行人目标进行散射强度阈值范围训练。
利用多次外场试验，获取行人目标在不同距离处的散射强度，对该散射强度取平均，进而获取散射强度阈值查找表。并根据多次测量获取能量起伏范围，例如±P_a，其中P_a为常数。根据检测目标的距离选取散射强度阈值，若检测目标的散射强度处于设定门限±P_a范围内，则满足行人目标的散射强度条件。
(2)对行人目标设定速度阈值范围。
设定行人目标的速度不超过10m/s，该速度范围可调。若检测目标的速度超出该速度范围，则认为该目标不是行人目标。
(3)对行人目标进行多普勒能量阈值范围训练
通过多次训练选取最为合适的多普勒能量阈值，依据雷达系统噪声信号强度确定多普勒能量阈值范围。具体的，统计非目标的平均多普勒能量，例如结果为P_n，选取多普勒能量阈值+P_n作为多普勒能量阈值范围，其中P_n为常数。
(4)对行人目标进行多普勒单元数目阈值训练。
当检测目标周围的多普勒单元对应强度满足(3)中设定的多普勒能量阈值范围，则计数加1，当计数超过多普勒单元数目阈值时，则满足行人目标的多普勒特征，即满足多普勒判决条件。该多普勒单元数目阈值通过多个行人目标统计取均值获取。
(5)对检测目标进行行人概率计算。
雷达对检测目标探测一次，获取一次探测数据，确定一次目标为行人概率。即，每探测一次获取一帧探测数据，确定一次目标为行人的可信度。
若对于单帧数据，检测目标同时满足行人散射强度阈值范围、行人速度阈值范围及多普勒判决条件，则检测目标的行人概率Prob_i增加概率Prob1；若目标不满足强度判决，则行人概率Prob_i减去概率Prob2；若目标不满足速度判决，则行人概率Prob_i减去概率Prob3，其中，Prob1，Prob2，Prob3为常数，根据多个行人目标训练获得。
(6)设定行人检测概率门限
a)通过单帧数据对检测目标进行确定是否是行人目标，由于探测过程中可能受到干扰，因此会出现大量的误判。因此这里通过多帧累积联合判断，即通过多次探测同一检测目标，根据检测目标的当前探测结果以及检测目标的上一次可信度，确定检测目标的当前可信度；并根据当前可信度，确定所述检测目标是否为行人目标。在设定的帧数范围内，当检测目标的当前可信度大于第一预设可信度门限，则将该目标判断为行人目标。若帧数达到设定的帧数，当检测目标的当前可信度仍未达到第一预设可信度门限，则将该目标判断为非行人目标。
b)为了避免行人检测结果出现抖动现象，也即当前帧判断检测目标为行人目标，下一帧判断为非行人目标，紧接着又检测为行人目标。本发明设定行人检测退出迟滞判断，在当前帧判断检测目标为行人目标后，将第一预设可信度更新为较小的值，并继续对下一帧进行判断。在后续对检测目标进行探测、获取探测数据、确定当前可信度的过程中，在有限的探测次数内，例如第二预设探测次数内，一旦确定检测目标的当前可信度小于更新后的第一预设可信度，则将检测目标确定为非行人目标；若在第二预设探测次数内，每次确定检测目标的可信度大于等于更新后的第一预设可信度，则将检测目标确定为行人目标。
当下一帧确定的检测目标的当前可信度低于设定可信度门限时，才由行人检测状态变为非行人检测状态。该门限设置数值比a)中门限小，即将第一预设可信度更新为较小的值。
该具体实施例示出了对检测目标确定是否是行人目标。应理解，对于具有微多普勒特征的检测目标，都适用于通过本实施例所述的技术方案对检测目标进行目标类型的确定。例如，根据无人机的螺旋桨运动时具有微多普勒特征，可以确定检测目标是否为无人机。
在图2至图5所示的实施例的基础上，本申请实施例提供了一种目标可信度确定系统，图9为本申请实施例提供的一种目标可信度确定系统的结构示意图。如图9所示，目标可信度确定系统90包括：处理器901。可选的，处理器901为一个或多个，共同地或单独地工作，所述处理器901用于执行上述目标可信度确定方法的实施例的技术方案。
可选的，目标可信度确定系统90还包括：存储器902、雷达903。 所述存储器用于存储程序代码；
所述处理器901，调用所述程序代码，当程序代码被执行时，用于执行以下操作：
获取检测目标的检测信息以及所述检测目标周围的多普勒单元的信息；
根据所述检测目标的所述检测信息以及所述检测目标周围的多普勒单元的信息，确定所述检测目标的类型的可信度。
可选的，雷达903与处理器901分立设置。可选的，雷达903包括处理器901。
本实施例提供的目标可信度确定系统能够执行上述目标可信度确定方法的实施例的技术方案，且执行方式和有益效果类似，在这里不再赘述。
在图6至图7所示的实施例的基础上，本申请实施例提供了一种目标识别系统，图10为本申请实施例提供的一种目标识别系统的结构示意图，如图10所示，目标识别系统100包括：目标识别系统100包括：处理器1001。可选的，处理器1001为一个或多个，共同地或单独地工作，所述处理器用于执行上述目标识别方法的实施例的技术方案。
可选的，目标识别系统100还包括：存储器1002以及雷达1003。所述存储器1002用于存储程序代码；
所述处理器1001，调用所述程序代码，当程序代码被执行时，用于执行以下操作：
多次探测同一检测目标；
根据所述检测目标的当前探测结果以及所述检测目标的上一次可信度，确定所述检测目标的当前可信度；
根据所述当前可信度，确定所述检测目标为特定的目标类型。
可选的，雷达1003与处理器1001分立设置。可选的，雷达1003包括处理器1001。
本实施例提供的目标识别系统系统能够执行上述目标识别系统方法的实施例的技术方案，且执行方式和有益效果类似，在这里不再赘述。
在上述实施例的基础上，本申请实施例提供了一种雷达。图11为本申请实施例提供的一种雷达的结构示意图，如图11所示，雷达 110包括：天线1101以及处理器1102。
天线1101，所述天线用于获取回波信号；
处理器1102，与所述天线通信连接，用于执行上述目标可信度确定方法实施例的技术方案，和/或，用于执行目标识别方法实施例的技术方案。
在图9所示的实施例的基础上，本申请实施例提供了一种可移动平台，图12为本申请实施例提供的一种可移动平台120，如图12所示，可移动平台120包括：机体1201、动力系统1202以及上述技术方案目标可信度确定系统90。
其中，可移动平台120可以是如下任意一种：机器人、无人飞行器、无人车、普通车辆、VR眼镜、AR眼镜。
图12所示实施例的可移动平台可用于执行上述目标可信度确定方法实施例的技术方案，其实现原理和技术效果类似，此处不再赘述。
在图10所示的实施例的基础上，本申请实施例提供了一种可移动平台，图13为本申请实施例提供的一种可移动平台130，如图13所示，可移动平台130包括：机身1301、动力系统1302以及上述技术方案目标识别系统100。
可移动平台130可以是如下任意一种：机器人、无人飞行器、无人车、普通车辆、VR眼镜、AR眼镜。
图13所示实施例的可移动平台可用于执行上述目标识别方法实施例的技术方案，其实现原理和技术效果类似，此处不再赘述。
另外，本实施例还提供一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行以实现上述实施例所述的目标可信度确定方法和/或目标识别方法。
在本发明所提供的几个实施例中，应该理解到，所揭露的装置、系统和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，所述单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分 开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外，在本发明各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用硬件加软件功能单元的形式实现。
上述以软件功能单元的形式实现的集成的单元，可以存储在一个计算机可读取存储介质中。上述软件功能单元存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)或处理器(processor)执行本发明各个实施例所述方法的部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(Read-Only Memory，ROM)、随机存取存储器(Random Access Memory，RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
本领域技术人员可以清楚地了解到，为描述的方便和简洁，仅以上述各功能模块的划分进行举例说明，实际应用中，可以根据需要而将上述功能分配由不同的功能模块完成，即将装置的内部结构划分成不同的功能模块，以完成以上描述的全部或者部分功能。上述描述的装置的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
最后应说明的是：以上各实施例仅用以说明本发明的技术方案，而非对其限制；尽管参照前述各实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分或者全部技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本发明各实施例技术方案的范围。

Claims (46)

  1. 一种目标可信度确定方法，应用于雷达，其特征在于，包括：
    获取检测目标的检测信息以及所述检测目标周围的多普勒单元的信息；
    根据所述检测目标的所述检测信息以及所述检测目标周围的多普勒单元的信息，确定所述检测目标的类型的可信度。
  2. 根据权利要求1所述的方法，其特征在于，所述检测目标的所述检测信息包括如下至少一种：距离、散射强度、速度信息。
  3. 根据权利要求1所述的方法，其特征在于，在所述检测目标的所述检测信息满足预设条件，且所述检测目标周围的多普勒单元的信息满足预设多普勒单元条件时，所述可信度增加第一预设值；否则，所述可信度减去第二预设值。
  4. 根据权利要求3所述的方法，其特征在于，所述检测目标的所述检测信息包括散射强度以及速度信息，
    其中，当所述速度信息在预设速度阈值范围内，所述散射强度与预设散射强度阈值范围内时，确定所述检测目标的所述检测信息满足预设条件。
  5. 根据权利要求3所述的方法，其特征在于，所述检测目标的所述检测信息包括散射强度，
    所述散射强度在预设散射强度阈值范围内时，确定所述检测目标的所述检测信息满足预设条件。
  6. 根据权利要求3所述的方法，其特征在于，所述检测目标的所述检测信息包括速度信息，
    所述速度信息在预设速度阈值范围内时，确定所述检测目标的所述检测信息满足预设条件。
  7. 根据权利要求4或5所述的方法，其特征在于，所述获取所述检测目标的检测信息之前，还包括：
    获取外部目标在不同距离处的所述散射强度；
    根据所述获取外部目标在所述不同距离处的所述散射强度，确定所述散射强度阈值范围。
  8. 根据权利要求1所述的方法，其特征在于，
    在所述检测目标的检测信息满足第一预设条件时，则获取所述检测目标周围的多普勒单元的信息。
  9. 根据权利要求8所述的方法，其特征在于，具体包括：
    所述检测信息包括距离以及散射强度，若所述散射强度在预设散射强度阈值范围内，和/或，所述检测信息包括速度信息，若所述速度信息在预设速度阈值范围内，则所述检测目标的检测信息满足第一预设条件；否则，所述检测目标的检测信息不满足所述第一预设条件。
  10. 根据权利要求1所述的方法，其特征在于，
    在获取所述检测目标的检测信息的同时，获取所述检测目标周围的多普勒单元的信息。
  11. 根据权利要求1所述的方法，其特征在于，包括：
    所述检测目标周围的多普勒单元的信息包括所述检测目标周围的每个多普勒单元的多普勒能量、以及所述散射强度在预设散射强度阈值范围内的所述多普勒单元的数目；
    根据所述散射强度在预设散射强度阈值范围内的所述多普勒单元的数目确定所述可信度。
  12. 根据权利要求11所述的方法，其特征在于，包括：
    获取所述散射强度在预设散射强度阈值范围内的所述多普勒单元的数目；
    根据所述多普勒单元的数目，确定所述多普勒单元的数目是否在预设多普勒单元数目范围内，并确定所述可信度。
  13. 根据权利要求12所述的方法，其特征在于，包括：
    获取所述检测目标周围的每个多普勒单元的多普勒能量；
    根据所述每个多普勒单元的所述多普勒能量，确定所述散射强度在预设散射强度阈值范围内的所述多普勒单元的数目。
  14. 根据权利要求11所述的方法，其特征在于，包括：
    若所述多普勒单元的数目在预设多普勒单元数目范围内，则所述可信度增加第三预设值；和/或，
    若所述多普勒单元的数目不在所述预设多普勒单元数目范围内，则所述可信度减去第四预设值。
  15. 根据权利要求1所述的方法，其特征在于，还包括：
    获取所述距离信息；
    根据所述距离信息，判断所述距离信息是否小于预设距离；
    若所述距离信息小于预设距离，则控制所述可移动平台执行避障操作，和/或，控制报警装置进行报警处理。
  16. 根据权利要求1所述的方法，其特征在于
    通过雷达获取所述检测目标的检测信息。
  17. 根据权利要求16所述的方法，其特征在于，所述雷达为毫米波雷达。
  18. 根据权利要求1所述的方法，其特征在于，通过获取回波信号，获取所述检测目标的检测信息。
  19. 一种目标识别方法，其特征在于，包括：
    多次探测同一检测目标；
    根据所述检测目标的当前探测结果以及所述检测目标的上一次可信度，确定所述检测目标的当前可信度；
    根据所述当前可信度，确定所述检测目标为特定的目标类型。
  20. 根据权利要求19所述的方法，其特征在于，每对所述同一检测目标进行探测一次，获取一次所述同一检测目标的所述探测结果。
  21. 根据权利要求19所述的方法，其特征在于，
    若所述当前探测结果满足第二预设条件，所述当前可信度为所述上一次可信度增加第一预设值；否则，所述当前可信度为所述上一次可信度减去第二预设值。
  22. 根据权利要求21所述的方法，其特征在于，所述检测目标的所述探测结果包括检测目标的检测信息、以及所述检测目标 周围的多普勒单元的信息；
    所述第二预设条件包括：当前时刻所述检测目标的检测信息满足预设检测信息条件以及所述检测目标周围的多普勒单元的信息满足预设多普勒条件。
  23. 根据权利要求19所述的方法，其特征在于，包括：
    根据对所述同一检测目标进行探测的次数，以及所述当前可信度确定所述检测目标的类型。
  24. 根据权利要求23所述的方法，其特征在于，当所述对所述同一检测目标进行探测的次数小于第一预设探测次数，将每次对所述同一检测目标进行探测确定的所述检测目标的当前可信度与第一预设可信度进行比较；
    若每次对所述同一检测目标进行探测确定的所述检测目标的当前可信度均小于所述第一预设可信度，则确定所述检测目标不是所述特定的目标类型；和/或，
    若一旦所述当前可信度大于等于所述第一预设可信度，则确定所述检测目标是所述特定的目标类型。
  25. 根据权利要求23所述的方法，其特征在于，当对所述同一检测目标进行探测的次数小于所述第一预设探测次数，所述当前可信度大于所述第一预设可信度，则第一预设可信度更新为第五预设值。
  26. 根据权利要求25所述的方法，其特征在于，当所述对所述同一检测目标进行探测的次数小于所述第一预设探测次数，所 述当前可信度大于所述第一预设可信度后，对所述检测目标进行继续探测，其中，在继续探测第二预设探测次数之前对所述检测目标的目标类型进行确定；
    若每次对所述同一检测目标进行探测确定的所述检测目标的当前可信度均大于等于第一预设可信度，则确定所述检测目标是所述特定的目标类型；和/或，
    对所述同一检测目标进行探测确定的所述检测目标的当前可信度均小于第一预设可信度，则确定所述检测目标不是所述特定的目标类型。
  27. 根据权利要求19所述的方法，其特征在于，当确定所述检测目标为特定的目标类型，则控制报警装置进行报警处理
  28. 根据权利要求19所述的方法，其特征在于，还包括：
    确定所述检测目标为特定的目标类型后，通过LED灯显示、语音播报、震动方式，输出所述检测目标的目标类型。
  29. 根据权利要求19所述的方法，其特征在于，通过雷达多次探测同一检测目标。
  30. 根据权利要求19所述的方法，其特征在于，所述雷达为毫米波雷达。
  31. 根据权利要求19所述的方法，其特征在于，根据回波信号获取所述检测目标的当前探测结果。
  32. 一种目标可信度确定系统，其特征在于，包括：一个或多个处理器，共同地或单独地工作，所述处理器用于执行如权利 要求1至15任一项所述的方法。
  33. 根据权利要求32所述的系统，其特征在于，通过雷达获取所述检测目标的检测信息。
  34. 根据权利要求32所述的系统，其特征在于，所述雷达为毫米波雷达。
  35. 根据权利要求32所述的系统，其特征在于，通过获取所述回波信号，获取所述检测目标的检测信息。
  36. 一种目标识别系统，其特征在于，包括：一个或多个处理器，共同地或单独地工作，所述处理器用于执行如权利要求19至27任一项所述的方法。
  37. 根据权利要求36所述的系统，其特征在于，通过雷达多次探测同一检测目标。
  38. 根据权利要求36所述的系统，其特征在于，所述雷达为毫米波雷达。
  39. 根据权利要求36所述的系统，其特征在于，通过获取所述回波信号，获取所述检测目标的检测信息。
  40. 一种雷达，其特征在于，包括：
    天线，所述天线用于获取回波信号；
    处理器，与所述天线通信连接，用于执行如权利要求1至15任一项所述的方法，和/或，用于执行如权利要求19至28任一项所述的方法。
  41. 根据权利要求40所述的雷达，其特征在于，所述雷达为 毫米波雷达。
  42. 根据权利要求40所述的雷达，其特征在于，通过获取所述回波信号，获取所述检测目标的检测信息。
  43. 一种可移动平台，其特征在于，包括：
    机体；
    动力系统，安装在所述车身，用于提供动力；
    以及权利要求40至42中任一项所述的雷达。
  44. 根据权利要求43所述的可移动平台，其特征在于，所述可移动平台至少包括如下一种：汽车、无人飞行器。
  45. 一种计算机可读存储介质，其特征在于，其上存储有计算机程序，所述计算机程序被处理器执行以实现权利要求1-18任一项所述的方法。
  46. 一种计算机可读存储介质，其特征在于，其上存储有计算机程序，所述计算机程序被处理器执行以实现权利要求19-31任一项所述的方法。
PCT/CN2019/073964 2019-01-30 2019-01-30 一种目标可信度确定方法、一种目标识别方法、系统、车辆及存储介质 WO2020154962A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2019/073964 WO2020154962A1 (zh) 2019-01-30 2019-01-30 一种目标可信度确定方法、一种目标识别方法、系统、车辆及存储介质
CN201980005757.0A CN111406224A (zh) 2019-01-30 2019-01-30 一种目标可信度确定方法、一种目标识别方法、系统、车辆及存储介质

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2019/073964 WO2020154962A1 (zh) 2019-01-30 2019-01-30 一种目标可信度确定方法、一种目标识别方法、系统、车辆及存储介质

Publications (1)

Publication Number Publication Date
WO2020154962A1 true WO2020154962A1 (zh) 2020-08-06

Family

ID=71437641

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/073964 WO2020154962A1 (zh) 2019-01-30 2019-01-30 一种目标可信度确定方法、一种目标识别方法、系统、车辆及存储介质

Country Status (2)

Country Link
CN (1) CN111406224A (zh)
WO (1) WO2020154962A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11662449B2 (en) * 2020-06-22 2023-05-30 Honeywell International Inc. Methods and systems for improving target detection performance of an indoor radar sensor

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111880160B (zh) * 2020-08-10 2023-01-31 深圳电目科技有限公司 基于雷达的人车识别方法及系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102156282A (zh) * 2011-03-25 2011-08-17 电子科技大学 一种基于微多普勒效应的雷达目标检测方法
US20140022118A1 (en) * 2009-11-03 2014-01-23 Vawd Applied Science And Technology Corporation Standoff range sense through obstruction radar system
CN105242254A (zh) * 2015-10-22 2016-01-13 中国船舶重工集团公司第七二四研究所 一种基于数据质量评估的对空目标识别方法
CN106990398A (zh) * 2016-01-21 2017-07-28 中国人民解放军空军工程大学 一种旋转对称目标微动特征认知提取方法

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8026842B2 (en) * 2006-06-08 2011-09-27 Vista Research, Inc. Method for surveillance to detect a land target
CN102866398B (zh) * 2012-09-21 2014-03-19 中国航天空气动力技术研究院 一种利用调频连续波雷达进行动目标识别的方法及系统
CN104937436B (zh) * 2013-02-01 2017-04-26 三菱电机株式会社 雷达装置
CN104215952B (zh) * 2014-08-27 2017-06-06 苏州闻捷传感技术有限公司 基于微动特性的车载目标识别系统及其识别方法
CN106405556B (zh) * 2016-11-02 2019-03-08 上海神添实业有限公司 车辆目标信息探测识别系统及其信号处理方法
CN107515395A (zh) * 2017-08-24 2017-12-26 北京航空航天大学 一种基于频率步进信号的近距离目标探测雷达

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140022118A1 (en) * 2009-11-03 2014-01-23 Vawd Applied Science And Technology Corporation Standoff range sense through obstruction radar system
CN102156282A (zh) * 2011-03-25 2011-08-17 电子科技大学 一种基于微多普勒效应的雷达目标检测方法
CN105242254A (zh) * 2015-10-22 2016-01-13 中国船舶重工集团公司第七二四研究所 一种基于数据质量评估的对空目标识别方法
CN106990398A (zh) * 2016-01-21 2017-07-28 中国人民解放军空军工程大学 一种旋转对称目标微动特征认知提取方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
GUAN, YUE: "Study on Methods of Target Recognition with Vehicle Millimeter Wave Radar", MASTER THESIS, 30 June 2018 (2018-06-30), pages 1 - 86, XP009522384 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11662449B2 (en) * 2020-06-22 2023-05-30 Honeywell International Inc. Methods and systems for improving target detection performance of an indoor radar sensor

Also Published As

Publication number Publication date
CN111406224A (zh) 2020-07-10

Similar Documents

Publication Publication Date Title
US10451720B2 (en) Object detection apparatus and object detection method
CN102765365B (zh) 基于机器视觉的行人检测方法及行人防撞预警系统
JP2020091281A (ja) レーダーデータを処理する装置及び方法
CN109814080B (zh) 一种毫米波雷达目标跟踪与置信度算法及其装置
CN105620489A (zh) 驾驶辅助系统及车辆实时预警提醒方法
CN110816540B (zh) 交通拥堵的确定方法、装置、系统及车辆
CN104237881A (zh) Fmcw防碰撞雷达多目标检测与跟踪系统及方法
JP3520326B2 (ja) ミリ波レーダによる走行車両検出方法
CN111123262B (zh) 自动驾驶3d建模方法、装置及系统
JP6557923B2 (ja) 車載レーダ装置及び領域検出方法
WO2020154962A1 (zh) 一种目标可信度确定方法、一种目标识别方法、系统、车辆及存储介质
CN104407346A (zh) 一种基于信息融合的移动式跑道异物监测方法
CN115034324B (zh) 一种多传感器融合感知效能增强方法
CN113156417B (zh) 反无人机探测系统、方法和雷达设备
CN110320517A (zh) 一种车载雷达识别方法
CN115257784A (zh) 基于4d毫米波雷达的车路协同系统
CN111077539B (zh) 基于激光雷达的探鸟系统
CN110422173B (zh) 一种行车环境识别方法
Yang et al. On-road collision warning based on multiple FOE segmentation using a dashboard camera
CN202130447U (zh) 一种新型的车道线偏离检测装置
CN116299424B (zh) 一种无人机识别系统及方法
CN111098890B (zh) 一种基于毫米波雷达的列车防撞方法及装置
CN105730330B (zh) 行车安全系统及其障碍物筛选方法
CN107015228A (zh) 车用毫米波雷达的环境辨识系统
CN116660934A (zh) 车辆、雨/雪天气识别方法、车辆控制方法和系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19912371

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19912371

Country of ref document: EP

Kind code of ref document: A1